This action might not be possible to undo. Are you sure you want to continue?

Welcome to Scribd! Start your free trial and access books, documents and more.Find out more

**An Option Pricing Approach
**

Werner Rosenberger

7

7

0

0

+

+

D

D

V

V

D

D

’

’

s

s

F

F

O

O

R

R

S

S

A

A

L

L

E

E

&

&

E

E

X

X

C

C

H

H

A

A

N

N

G

G

E

E

w ww ww w. .t tr ra ad de er rs s- -s so of ft tw wa ar re e. .c co om m

w ww ww w. .f fo or re ex x- -w wa ar re ez z. .c co om m

w ww ww w. .t tr ra ad di in ng g- -s so of ft tw wa ar re e- -c co ol ll le ec ct ti io on n. .c co om m

w ww ww w. .t tr ra ad de es st ta at ti io on n- -d do ow wn nl lo oa ad d- -f fr re ee e. .c co om m

C Co on nt ta ac ct ts s

a an nd dr re ey yb bb br rv v@ @g gm ma ai il l. .c co om m

a an nd dr re ey yb bb br rv v@ @y ya an nd de ex x. .r ru u

S Sk ky yp pe e: : a an nd dr re ey yb bb br rv v

Risk-adjusted Lending Conditions

Wiley Finance Series

Measuring Market Risk

Kevin Dowd

An Introduction to Market Risk Measurement

Kevin Dowd

Behavioual Finance

James Montier

Asset Management: Equities Demystiﬁed

Shanta Acharya

An Introduction to Capital Markets: Products, Strategies Participants

Andrew Chisholm

Hedge Funds: Myths and Limits

Francois-Serge Lhabitant

The Manager’s Concise Guide to Risk

Jihad S Nader

Securities Operations: A Guide to Trade and Position Management

Michael Simmons

Modelling, Measuring and Hedging Operational Risk

Marcelo Cruz

Monte Carlo Methods in Finance

Peter J¨ ackel

Building and Using Dynamic Interest Rate Models

Ken Kortanek and Vladimir Medvedev

Structured Equity Derivatives: The Deﬁnitive Guide to Exotic Options and Structured Notes

Harry Kat

Advanced Modelling in Finance Using Excel and VBA

Mary Jackson and Mike Staunton

Operational Risk: Measurement and Modelling

Jack King

Advance Credit Risk Analysis: Financial Approaches and Mathematical Models to Assess, Price and

Manage Credit Risk

Didier Cossin and Hugues Pirotte

Dictionary of Financial Engineering

John F. Marshall

Pricing Financial Derivatives: The Finite Difference Method

Domingo A Tavella and Curt Randall

Interest Rate Modelling

Jessica James and Nick Webber

Handbook of Hybrid Instruments: Convertible Bonds, Preferred Shares, Lyons, ELKS, DECS and Other

Mandatory Convertible Notes

Izzy Nelken (ed)

Options on Foreign Exchange, Revised Edition

David F DeRosa

Volatility and Correlation in the Pricing of Equity, FX and Interest-Rate Options

Riccardo Rebonato

Risk Management and Analysis vol. 1: Measuring and Modelling Financial Risk

Carol Alexander (ed)

Risk Management and Analysis vol. 2: New Markets and Products

Carol Alexander (ed)

Interest-Rate Option Models: Understanding, Analysing and Using Models for Exotic Interest-Rate

Options (second edition)

Riccardo Rebonato

Risk-adjusted Lending Conditions

An Option Pricing Approach

Werner Rosenberger

Published 2003 by John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester,

West Sussex PO19 8SQ, England

Telephone (+44) 1243 779777

Copyright

C

2003 Werner Robert Rosenberger

First edition published in German by Paul Haupt Verlag, Berne, Switzerland in paperback in 2000

Translated by Christopher Massy-Beresford, 4 Vaughan Avenue, London W6 0XS

Email (for orders and customer service enquiries): cs-books@wiley.co.uk

Visit our Home Page on www.wiley.co.uk or www.wiley.com

All Rights Reserved. No part of this publication may be reproduced, stored in a retrieval system or

transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning or

otherwise, except under the terms of the Copyright, Designs and Patents Act 1988 or under the terms of

a licence issued by the Copyright Licensing Agency Ltd, 90 Tottenham Court Road, London W1P 0LP,

UK, without the permission in writing of the Publisher. Requests to the Publisher should be addressed to

the Permissions Department, John Wiley & Sons Ltd, Bafﬁns Lane, Chichester, West Sunsex PO19

1UD, England or emailed to permreq@wiley.co.uk, or faxed to (+44) 1243 770620.

This publication is designed to provide accurate and authoritative information in regard to the subject

matter covered. It is sold on the understanding that the Publisher is not engaged in rendering

professional services. If professional advice or other expert assistance is required, the services of a

competent professional should be sought.

Other Wiley Editorial Ofﬁces

John Wiley & Sons Inc., 111 River Street, Hoboken, NJ 07030, USA

Jossey-Bass, 989 Market Street, San Francisco, CA 94103-1741, USA

Wiley-VCH Verlag GmbH, Pappelallee 3, D-69469 Weinheim, Germany

John Wiley & Sons Australia Ltd, 33 Park Road, Milton, Queensland 4064, Australia

John Wiley & Sons (Asia) Pte Ltd, 2 Clementi Loop #02-01, Jin Xing Distripark, Singapore 129809

John Wiley & Sons Canada Ltd, 22 Worcester Road, Etobicoke, Ontario, Canada M9W 1L1

Wiley also publishes its books in a variety of electronic formats. Some content that appears in print

may not be available in electronic format.

Library of Congress Cataloging-in-Publication Data

Rosenberger, Werner.

[Risikoad¨ aquate Kreditkonditionen. English]

Risk-adjusted lending conditions : an option pricing approach / Werner Rosenberger ;

[translated by Christopher Massy-Beresford].

p. cm. —(Wiley ﬁnance series)

Includes bibliographical references and index.

ISBN 0-470-84752-2 (alk. paper)

1. Option (Finance) —Prices —Mathematical models.

2. Credit —Management —Mathematical models. 3. Risk management —Mathematical

models. I. Title. II. Series.

HG6024.A3 R67 2002

332.1

753

**0681 —dc21 2002031125
**

British Library Cataloguing in Publication Data

A catalogue record for this book is available from the British Library

ISBN 0-470-84752-2

Typeset in 10/12pt Times by TechBooks, New Delhi, India

Printed and bound in Great Britain by Biddles Ltd, Guildford and King’s Lynn

This book is printed on acid-free paper responsibly manufactured from sustainable forestry

in which at least two trees are planted for each one used for paper production.

For my family

Contents

Preface 1 xiii

Preface 2 xv

Part I Outline 1

1 Introduction 3

1.1 The problem 3

1.2 Narrowing the subject down, setting the objective and subdividing it 3

1.3 The insurance concept 5

1.4 Types of problem in the context of loan business 7

1.5 Loan interest rate model 8

1.6 Model for calculating risk surcharge 9

1.7 Assumptions 11

1.8 Testing the model 12

1.9 Loan exposure models 13

1.9.1 Classical methods 14

1.9.2 Modern credit risk analysis based on ﬁnancial theory 14

1.9.3 The model presented here seen in relation to previous models 16

2 Rating system 19

2.1 The need for a rating system 19

2.2 Deﬁning shortfall risk in terms of ﬁgures 20

2.3 Deﬁning the credit-worthiness key ﬁgure 21

2.4 Example of a rating system in terms of ﬁgures 21

2.5 Ampliﬁcation of the rating system for very competitive markets 23

viii Contents

Part II Mathematical Foundations of the Model 25

3 Probability model: Development of ψ

j

27

3.1 Determining the probability of cash ﬂows being fulﬁlled 27

3.2 Maturity transformation 28

3.3 Conclusions 29

3.3.1 The case of a loan being granted indeﬁnitely 29

3.3.2 Reﬂections on the success chance ε(n) 29

3.4 Results and conclusions 31

4 Calculation of the shortfall risk hedging rate in the special case

of shortfall risks being constant 33

4.1 Fixed advance without repayments 33

4.2 Fixed advance with regular repayments 36

4.3 Loans on regular annual repayment 36

4.4 Current account credit 36

4.5 Loan assessment 40

4.6 Conclusions 41

4.6.1 Minimum loan interest rate 42

4.6.2 Effective proﬁt contribution rate 43

4.6.3 Effective shortfall risk hedging rate 43

4.6.4 Maximum shortfall risk covered 44

4.7 Results and conclusions 45

4.8 Example 45

5 Calculation of the shortfall risk hedging rate in the general case

of variable shortfall risk 47

5.1 Fixed interest loan without repayments 47

5.2 Approximate solution for ﬁxed interest loan without repayments 48

5.3 Reliability of the approximate solution 49

5.4 Fixed advance with complete repayment 50

5.5 Fixed advance with partial repayments 51

5.6 Current account loans 52

5.7 Results and conclusions 52

6 Shortfall risk on uncovered loans on the basis of statistics 53

6.1 Private clients 53

6.1.1 Unearned income and income from self-employment 53

6.1.2 Income from salaried employment 54

6.1.3 Investment income and assets 54

6.2 Companies 55

Contents ix

Part III Option-Theory Loan Risk Model 57

7 Shortfall risk on uncovered loans to companies on the basis of an

option-theory approach 59

7.1 Difference in approach between Black/Scholes and KMV, together

with further elaboration 59

7.2 Derivation of basic formulae 60

7.3 Derivation of risk-adjusted values 63

7.4 Deﬁnition of the values for the solution formula 67

7.4.1 The value of the company and its debt rate 67

7.4.2 Volatility 69

7.4.3 Private debtors 71

7.5 Inﬂuence of individual parameters on the credit shortfall risk 71

7.6 Risk of bankruptcy and breakdown distribution 78

7.7 Loan assessment 79

7.8 Bonds 80

7.9 Consideration of privileged salary claims in the event of bankruptcy 80

7.10 Limits to the application of the option theory approach 82

7.11 Results and conclusions 84

7.12 Examples 85

7.12.1 Example of a company with continuous business development 85

7.12.2 Example of a company with a poor ﬁnancial year 87

8 Loans covered against shortfall risk 91

8.1 Shortfall risk of a covered loan on the basis of the option-theory

approach 91

8.2 Correlation between the shortfall risk of the borrower and the shortfall

risk of the collateral 93

8.2.1 Derivation of the correlation 93

8.2.2 Value area of the efﬁciency of the correlation 94

8.3 Shortfall risk of the covered loan 97

8.4 Covered and uncovered loans to the same borrower 98

8.5 Results and conclusions 99

8.6 Example 99

9 Calculation of the combination of loans with the lowest interest costs 103

9.1 Marginal interest rate 103

9.2 Two loans 105

9.3 Three loans 107

9.4 The general case of several loans 107

9.5 Partially covered loans 108

x Contents

9.6 Maximum return on equity 108

9.7 Acceptability of debt servicing 112

9.7.1 Acceptability of interest rates 112

9.7.2 Acceptability of repayment 113

9.7.3 Maximum debt 114

9.7.4 Consequences for companies with declining earning 114

9.7.5 Consequences for loan supervision 114

9.8 Results and conclusions 115

9.9 Example 116

Part IV Implementation in practice 119

10 Procedure —according to the model —for assessing the risk

in lending to a company 121

10.1 Overall view of the procedure 121

10.2 Analysis of earnings statements 122

10.3 Analysis of cash ﬂow statements 122

10.4 Analysis of balance sheets 123

10.5 Determination of the discount rate 123

10.6 Determination of relevant loan terms 124

10.7 Determination of volatility 124

10.8 Determination of credit risk 124

10.9 Prudence in the case of new loans/borrowers 125

10.10 Possible causes of conﬂict between bank and borrower when the

model is applied 125

11 Applications 127

11.1 Specialist clothing business: Turn-around situation 127

11.2 Company trading in machine tools: Provision for successor company 129

11.3 Ship mortgages: Risk limitation 130

11.3.1 Starting position 130

11.3.2 The banks’ loan decision 130

11.3.3 Assessment of the loan risk by the banks 131

11.3.4 Determination of loan risk according to the model 131

11.3.5 Comparison of assessment between the bank and the model 134

11.3.6 Limitation of loan risk 134

11.4 Mortgage business 1985–99 135

12 Final considerations 139

12.1 Tests of hypotheses 139

12.2 Implementation in practice 141

12.2.1 Specialist personnel 141

12.2.2 Statistical bases 142

Contents xi

12.2.3 IT support 143

12.2.4 Organisational measures 143

12.3 Applicability of the method presented 144

12.4 Customer considerations 145

12.5 Open questions 145

12.6 Closing remarks 146

Appendix 1: Notation 147

Appendix 2: Excel worksheet 149

Appendix 3: Property price index 151

Appendix 4: Chapter 3 —Derivations 153

Appendix 5: Chapter 4 —Derivations 155

Appendix 6: Chapter 5 —Derivations 159

Bibliography 163

Index 167

Preface 1

It is unusual to tackle a dissertation project 14 years after completing one’s studies. In my

case it was an assignment at the Swiss Banking School that triggered it off. Professor Dr Ernst

Kilgus, who was its Director at the time, encouraged me to develop my assignment, and we

agreed in the course of our conversations to bring this about in the form of a dissertation.

At that point I had seriously underestimated the time that would be needed for this under-

taking. I had already been a member of Credit Suisse’s management team for some years, and

was neither able nor willing to work there only part time. So that is the main reason why it

has taken ﬁve years for this book to emerge. Looking back, however, I can certainly put on

record howpleased I am, on the strength of howit has turned out, to have brought this study into

being.

I had to reduce the length of the ﬁrst plan considerably, in order to be able to keep the scope

of this dissertation manageable from the view of the reader’s time as well as my own. It was

originally envisaged that calculations of loan derivatives and assessment of a company’s equity

would be examined using the methods described here. Although initial results have already

been obtained on these subjects, these have had to be left out —but they will form the subject

of later publications.

Professor Kilgus said on the basis of our discussion that he was prepared to act as my super-

visor for this project. Thanks to his critical observations this book nowcontains many practical

examples and advice to the reader, with the objective of making this work comprehensible and

useful to a wider public. I am extremely grateful to my supervisor for his thought-provoking

comments, which compelled me to question things again and again. Without him this book

would have been substantially more theoretical and thus less easy to read and understand.

On the academic side I also owe many thanks to Professor Diethard Klatte and R¨ udiger Frey

of the University of Z¨ urich and to Dr Philipp Halbherr of the Cantonal Bank of Z¨ urich. They

had declared their willingness to submit my manuscript to critical examination, and provided

me with many valuable ideas that have contributed to the success of this study.

I thank my employer, Credit Suisse, for its indulgence in allowing me to attend some of the

normal lectures at the University of Z¨ urich during working hours. Included especially in this

vote of thanks are many of my colleagues, with whomI was able to have interesting discussions

on the subject and who helped in working up the examples.

xiv Preface 1

When it came to this book, I was able to count on the one hand on the considerable support

of Mr H. P. Wyssmann, who scoured it with meticulous dependability for mistakes and inac-

curacies; and on the other, I managed to enlist the assistance of my brother, J¨ urg Rosenberger,

in its design and layout. Both have my sincere thanks for their invaluable help.

My greatest thanks of all go to my partner Beatrice Wyssmann for the constant exchange

of ideas, which covered absolutely every aspect of the work, from the simplest to the most

complicated, and for her patience over its ﬁve-year gestation. Thanks to her support I was able

to bring it to fruition in a calm and orderly atmosphere.

May 2000

Walchwil

Switzerland Werner R. Rosenberger

Preface 2

Books about topics in modern ﬁnance theory have to be published in English today in order

to reach an international audience. This is why I always wanted to publish my book in

English too.

I met Sally Smith, Senior Publishing Editor at John Wiley &Sons, at a conference in Geneva

in the autumn of 2000. She wanted to publish my book right from the beginning and was of

great help to me at all times. I thank her and her successor, Samantha Whittaker, for all the

support they provided me.

The next step was to ﬁnd a translator. Stuart Benzie from McKinsey & Company, with

whom I worked on a project in London during that time, helped me to ﬁnd Christopher Massy-

Beresford for this job. I thank Christopher for his excellent work.

My special thanks go to my friend Carolina Schwyn-Villalaz. Her interest and her support

have always been very encouraging and helpful.

September 2002

Walchwil

Switzerland Werner R. Rosenberger

Part I

Outline Outline

Introduction

Rating system

1

Introduction

1.1 THE PROBLEM

A bank can only pursue lending business proﬁtably if all the costs of providing loans can

be covered by the income they produce. The bank’s income depends on the prices for loans

(see Section 1.5) that are charged to its borrowers. It is necessary to know the costs of each

loan concerned precisely in order to be able to charge a price that fully covers its costs.

An important element in the costs of making loans arises from the risks that are inseparably

bound up with such business. These risks, according to Kilgus [KILG94, S. 66ff], comprise:

r

shortfall risks

r

market risks

r

liquidity risks

r

behavioural risks

r

operational risks

Kilgus [KILG94, S. 69] asserts that it is absolutely essential that these risks, both in the

case of each individual loan and at the level of the bank’s operations as a whole, are identiﬁed,

assessed, controlled and supervised in a comprehensive risk management plan.

It is therefore necessary to be able, among other things, to assess these risks correctly

for calculating the cost-covering price of any loan —but achieving this objective is far from

simple (cf. [DRZ198]). The necessity, however, also results from the fact, as Berger and

DeYoung [BEDE97] succeeded in showing, that a high number of troubled loans brings about

a signiﬁcant deterioration in cost efﬁciency. The greater the number of troubled loans that

have to be managed, the higher the administrative costs owing to increased expenditure on

supervision and winding up become (see also Section 1.5: proﬁt contribution rate).

The more precisely the bank knows its risks exposure, the more precisely can it —in the same

way as insurance companies (cf. Section 1.3) —set up appropriate precautionary reserves.

Unforeseen provisions and write-offs owing to surprise losses on lending business, which

inevitably leave the bank’s shareholders with a poor impression, can thereby be largely avoided

(cf. [SDHJ97].

1.2 NARROWING THE SUBJECT DOWN, SETTING THE

OBJECTIVE AND SUBDIVIDING IT

This study will concentrate only on the domestic shortfall risks of any bank lending, in its

local currency, to businesses and private individuals, and will not cover lending to the public

sector. It will be assumed that the costs of any loan, including the costs that derive from the

other risks listed above, are known.

4 Risk-adjusted Lending Conditions

The objective is to develop a model that permits calculation of the lending costs which are

caused by the shortfall risk. Following Kilgus [KILG94, S.69] by shortfall risk is meant, in

this case, the fact that the borrower is either incapable of meeting, or unwilling to meet, the

bank’s demands.

To attain this objective, we will fall back on the option-price theory approach of Black and

Scholes [BLSC73, S. 637ff]. Our expositions will thus link directly with their explanations

(see Section 7.1). These will demonstrate that the Black/Scholes approach can be extended to

reach a generalised solution that is applicable in real life.

It is therefore emphasised quite clearly at this point that this study is concerned with devel-

oping a mathematical model, which demonstrates the properties speciﬁed. As with all models

that are intended to be of relevance in practice, it is necessary to compare the model’s state-

ments with real life on the basis of empirical testing. Undertaking such empirical testing as

well would, however, by far exceed the scope of this study. All that can be done here is to make

a qualitative comparison between the results and conclusions of the model, and experience in

the lending business generally. This will take place in the context of the examples concerned.

The fundamental basis of the model is put forward in this ﬁrst chapter. The second chapter

will tackle the problem of inaccuracies in the basic data involved, which a priori impair the

accuracy of the model and make a rating system necessary.

In Chapters 3, 4 and 5 —with the aid of probability calculus —the fundamental correlation

between the risk-free rate of interest, the risk-adjusted rate of interest, the probability of

bankruptcy, the breakdown distribution probability value, and the probability of shortfall on

loan, will all be derived. In the course of this we shall proceed, in Chapter 4, on the unrealistic

assumption that the shortfall risk is constant over the lifetime of any loan. This assumption will

be dropped in Chapter 5, and it will be shown how the general case of non-constant shortfall

risks can be linked up with the conclusions drawn in Chapter 4. We chose this course of action

in order to show clearly how the conclusions were drawn. That shortfall risks are not in fact

constant over time is demonstrated in Section 7.5 (cf. Figures 7.10 and 7.11).

Chapter 6 gives an indication of how the interrelations derived in the preceding expositions

may be applied with the aid of facts determined statistically. But all we will be doing here is

mentioning, to complete the picture, the application of statistics as one possible way of solving

the problem.

Chapter 7 contains the crucial part of this study. Here we go back to the original idea of

Black and Scholes (cf. [BLSC73]) of describing and evaluating debts and equity capital with

the aid of their option-price theory. This idea was taken up later by Merten, Geske and others,

but never led to completely satisfactory results. The main problemhere was that the calculation

of the rate of interest consistent with risk was always dependent on the risk-free rate of interest

that was consistent with the time-scale. For this reason it has not hitherto been possible to

evaluate enterprises with complicated structures on the liabilities side of the balance sheet, as

far as risk is concerned, consistently. Thanks to the Nobel Prize won in 1997 by Black, Scholes

and Merten this approach is currently undergoing a minor renaissance [ZIMM98]. Related

approaches are also being applied to this (cf. for example [GREN96] or [BRVA97]. As already

mentioned, these considerations will be further developed here.

The KMV Corporation in San Francisco, California (see [VAS184] and [KAEL98]) is

adopting a similar approach. Under this the volatility of a company’s value is inferred via the

Introduction 5

volatility of the stock exchange prices of listed companies. Their model does not, however, use

the Black/Scholes equation, but pursues a similar solution approach, likewise proceeding from

a stochastic process. In the course of this study, by way of contrast, the volatility of capitalised

free cash ﬂows is used to assess the volatility of the company’s value using a model similar

to the Black/Scholes equation. Loans to companies not listed on stock exchanges may also be

assessed using it.

The expositions in Chapter 7 will show that a small detour leads to the objective. The risk-

free rate of interest and the rate of interest consistent with risk in the Black/Scholes equation

may, with the assistance of the conclusions from Chapters 3, 4 and 5, be suitably replaced

by the credit shortfall risk. As a result, for any given company, the credit shortfall risk may

be calculated depending on the volatility of the company’s market value, on the debt rate

in relation to the market value, and on the relevant term of any part of the total liabilities.

This calculation may be undertaken separately for any part of the liabilities. On the basis of

the various credit shortfall risks, the appropriate loan interest rate consistent with risk can

be calculated individually for each part of the liabilities. Here the liabilities side can be as

complicated as you like. With the aid of analogous conclusions, loans to private individuals

that are consistent with risk may also be calculated.

Chapter 8 will examine how a loan may be calculated on the basis of collateral and how the

combination of borrower and collateral that is consistent with risk may be calculated correctly.

Chapter 9 will demonstrate howloans with varied collateral may be combined to bring about

debt ﬁnancing that is optimally structured in relation to risk for any company.

To clarify the expositions so far, Chapter 10 will describe —in the manner of a cookery

book —how to proceed in assessing loan risk according to this model. The application of the

methods described will be elaborated in Chapter 11, with examples of loan transactions from

actual banking practice.

Chapter 12 will demonstrate to what extent the considerations applying to checking loans

ought to be adjusted to line up with current normal procedures, i.e., how the empirical test

mentioned of the methods presented here may be carried out, and what preconditions any bank

would have to fulﬁl for their introduction.

1.3 THE INSURANCE CONCEPT

Hitherto banks have endeavoured to keep shortfall risks in their lending business as low as

possible, with losses on loans being regarded as individual cases, occurring unsystematically,

that should be avoided as best as possible. It has therefore to be the aim of every bank, among

other things —according to Zellweger [ZELL83, S. 1] —to keep the risks arising in connection

with the exercise of individual banking transactions as small as possible. Which risks that result

from the business of commercial lending represent the greatest and most commonly occurring

ones thus becomes a matter of special importance to any banking institution. Meier [MEIE96,

S. 3] also makes it the objective of his work to make a contribution to reducing and limiting

shortfall risks.

Traditionally it was the loan ofﬁcial’s job to identify problems of customer credit-worthiness

in good time, and if necessary to call the loan in. This is always akin to walking a tight-rope,

with premature calling in on one side, which has losses of proﬁts on loans and angry customers

6 Risk-adjusted Lending Conditions

as consequences, and on the other side calling-in operations that are too late and are associated

with the resulting losses in receivables. It is in the very nature of things that any calling in of a

loan that is too late is, in every case, seen as failure because of the loss involved. But even in

the case of calling in a loan in good time there is always uncertainty as to whether it was really

justiﬁed in the circumstances. This is extremely unsatisfactory, whichever way you look at it.

We shall therefore adopt a fundamentally different approach here, involving the development

not of methods of limitation, but of methods of calculating shortfall risks. Losses on loans will

not be considered as individual instances to be avoided, but as costs that are, as they are in

the insurance business, calculable. As soon as losses on loans become calculable, it becomes

possible to charge an appropriate risk-adjusted price for each loan. It is then up to the borrower

to decide whether or not it is willing to pay the price demanded. This way the decision on

whether any loan materialises no longer lies with the loan assessment department of the bank,

but with the client. The client decides whether or not it is prepared to pay the shortfall risk

premium included in the price for the loan.

The task of the bank’s loan assessment department therefore becomes fundamentally differ-

ent. Traditionally it had to decide whether or not the borrower’s ﬁnancial standing was up to

the granting of credit at prices that were more or less set in advance. The new task is to charge

the right shortfall risk premium for each and every loan, in much the same way as premiums

are calculated for insurance policies.

The result of this is that the department of the bank responsible for sales will only close

deals in which the borrower concerned is prepared to pay the price demanded. Here it is

perfectly legitimate, if applicable in view of the borrower’s other banking business, to grant

price concessions on loans, as long as the overall relationship with the client is proﬁtable. This

does, however, presuppose that skills in the bank’s accounting function have been appropri-

ately developed, and that it is in a position to prepare the information that will be needed.

It goes without saying that this course of action demands high standards of any bank’s sales

departments.

As could be seen in the Neue Z¨ urcher Zeitung [NZZ96, Nr. 218, S. 29/NZZ96, Nr. 298, S. 19]

the then Swiss Bank Corporation and the Credit Suisse Group have changed their working

methods for accruing liability reserves profoundly, with loan risks no longer representing

extraordinary occurrences, but nowbeing treated as calculable costs. Two major Swiss banks in

the lending business have already completed the move over to the insurance concept. According

to an article in the Schweizer Bank [SCM

¨

U98] it may also be assumed that the same is now

true for the Cantonal Bank of Z¨ urich.

Many banks today have become substantially more circumspect in their lending than they

were a few years back, because of high losses on loans in the ﬁrst half of the 1990s. Now

voices can be heard criticising this circumspection and reproaching the banks for exercising

excessive prudence. A study from the USA [CPSH98] comes to similar conclusions, in that

it establishes that ﬁnance companies there may be rather more inclined to take on higher

risks, when granting loans, than banks. In contrast to the USA there are, however, no ﬁnance

companies in Switzerland that would grant loans to companies to any extent worth mentioning.

As soon as credit risk costs can be calculated exactly, it will be possible for banks to take on

higher risks again ‘safely’. It is at any rate important here that the running of higher risks must

be associated with higher earnings. According to one study [DICH98] this does not appear

Introduction 7

to be the case in the USA. A comparable investigation of the situation in Switzerland has not

been undertaken.

1.4 TYPES OF PROBLEM IN THE CONTEXT OF LOAN BUSINESS

Our arguments so far rest on the assumption that one has to distinguish, in the lending business,

between the following types of problem. It is not absolutely necessary for this that the individual

types of problem be dealt with separately by different departments within the bank. It is,

however, interesting to note that this is the case among the banks in the Credit Suisse Group.

These types of problem are:

At the Individual Transaction Level

r

Calculationof the costs of reﬁnancingandunderpinninga loanfromownresources, including

the costs of liquidity and market risks (cf. [KILG94, S. 66ff]): a bank’s treasury department

usually undertakes this task.

r

Calculation of the running costs for the processing of a loan, including the costs of be-

havioural and operational risks (cf. [KILG94, S. 66ff]: a bank’s accounting department

usually undertakes this task.

r

Calculation of the shortfall risk costs of a loan: this is a question of a new task for the loan

assessment departments of banks.

r

Negotiation of price with the borrower, taking the overall proﬁtability of the customer into

account. The departments of a bank that are responsible for its customers normally undertake

this task.

r

Handling of delinquent loan business: this task is frequently undertaken by the loan assess-

ment departments of a bank. In the case of the Credit Suisse Group, for instance, there are

specialist units within the organisation responsible for this.

At the Overall Banking Level

r

Preparation of a management information system for assessing total lendings, and for eval-

uating the work of the departments of the bank involved in the lending business described

above.

r

Calculation of the own resources requirement and/or additional liability reserves to cover

statistical ﬂuctuations as actual losses appear in any given total lendings situation. This is

essentially a question of the calculation of cluster risks and of the effects of diversiﬁcation.

The tasks at the overall banking level should be assigned to an ofﬁce that is responsible

for the overall supervision and control of total lendings. Banks should be led by the insurance

concept in the building up of these organisational structures as well. Any management infor-

mation system should thus be in the same situation, as is normal in the case of any insurance

company, to provide all this information: in other words there is, for example, no fundamental

difference between management information systems for an insurance company’s motor lia-

bility insurance and a bank’s mortgage business. Losses calculated in advance according to the

8 Risk-adjusted Lending Conditions

model and losses actually incurred must be compared with each other on the basis of various

parameters, and the parameters further scrutinised for their relevance.

It would go beyond the remit of this present study to go more closely into the organisation

of a management information system and into the calculation of the own resources that would

be needed, as already implicitly indicated in Section 1.2, from a business management point

of view. At this juncture the intention is simply to point out the need concerned, in order to

complete the picture.

1.5 LOAN INTEREST RATE MODEL

The price of a loan is normally expressed in the form of a loan interest rate. Exceptions are, for

example, leasing business and consumer credit, the price of which is expressed in the form of

regular monthly payments. These regular monthly payments are, however, likewise determined

from imputed interest rates. Thus a loan interest rate has to be calculated in the same way as

the price for the loan. Our expositions on this are based on the following loan interest rate

model (cf. also [SCM

¨

U98]):

i = f + p +r (1.1)

i = loan interest rate

f = ﬁnancing cost rate

p = proﬁt contribution rate

r = shortfall risk hedging rate

Loan Interest Rate

The loan interest rate is deﬁned here as the rate that is charged to the borrower as the price for

the loan, and which is indeed paid by the borrower. An integrated loan interest rate is assumed

here for the sake of simplicity. Swiss practice also involves charging ‘loan commissions’ to

the borrower in the case of loans on current account. The ‘loan commissions’ form an element

of i , in this model, as the division between the loan interest itself and the bank’s commission

may in principle be undertaken at will. This simpliﬁcation may therefore be undertaken for

the purposes of the model.

Financing Cost Rate

The ﬁnancing cost rate is deﬁned here as that rate, the revenue from which will cover the costs

of reﬁnancing, of statutory underpinning by own resources, and the costs of market risks and

liquidity risks.

1

Here it has been assumed that the bank’s treasury can calculate this rate on

each loan at a ﬁxed, given level.

These costs accumulate in proportion to the amount of loan that is taken up.

1

In Switzerland today uncovered loans to companies have to be underpinned by own resources to an average extent of 8%. That,

however, does not mean that 8% of each individual loan has to be underpinned, irrespective of the debtor’s ﬁnancial standing, as is

often to be observed in banking practice. It is rather that, in calculations internal to banks, loans to debtors of good ﬁnancial standing

may be underpinned with less of the bank’s own resources than loans to debtors of worse ﬁnancial standing, as long as on average the

underpinning is still at least 8%.

Introduction 9

Proﬁt Contribution Rate

The proﬁt contribution rate is deﬁned here as the rate that covers the costs of handling the

loan, the behavioural risks and the operational risks, and permits an appropriate proﬁt (cf. also

Section 4.6) to be earned on the loan. That the behavioural risks are reﬂected in the proﬁt

contribution rate is connected with the following assumptions: normally the borrower is only

deﬁned as a behavioural risk if it comes under sustained ﬁnancial pressure as a result of

debt servicing charges. Behavioural risks thus mostly surface only when the borrower has no

further borrowing power. This leads to the fact that behavioural risk and shortfall risk are almost

identical due to absence of borrowing power. Minor deviations resulting from this assumption,

together with the few instances of fraudulent raising of loans, should therefore be ascertained

by statistical methods and may consequently be portrayed as part of the proﬁt contribution

rate.

We make the assumption here that the ofﬁce in the bank responsible for sales has to obtain

the highest possible loan interest rate in negotiations with the borrower, and therefore the

highest possible proﬁt contribution rate: the revenue from the ﬁnancing cost rate results, after

deduction of the above-mentioned costs, which do not accumulate in proportion to the amount

of credit that is taken up, in the earnings from the loan concerned. In this a multi-stage proﬁt

contribution rate calculation may be brought to bear. It will be explained in Section 4.6.2 that

negative proﬁt contributions and earnings may also result from any loan. We will demonstrate

there how the effective proﬁt contribution rate is calculated, if the following variables are

quantiﬁed —the loan interest rate negotiated with the customer, the ﬁnancing cost rate and the

shortfall risk.

As a more recent study shows [WONG97], there is an optimum interest margin for any

risk-averse bank in consideration of credit-worthiness and market risks. This puts a ceiling

on any bank’s lending activities, in that only those transactions are entered into in which the

margin, after deduction of shortfall risk costs, permits a sufﬁcient proﬁt contribution.

In the case of advances made at increased credit-worthiness risk even higher proﬁt con-

tribution rates should be estimated because of the increased administrative expenditure (cf.

[BEDE97]). And in the case of advances with higher credit-worthiness risk, increased admin-

istrative costs therefore lead also to higher loan interest rates, in addition to increased shortfall

risk costs.

Shortfall Risk Hedging Rate

The shortfall risk hedging rate is deﬁned here as whatever cost rate covers the shortfall risk

costs of the loan concerned. We assume that this rate is dependent on the ﬁnancial standing of

the borrower.

1.6 MODEL FOR CALCULATING RISK SURCHARGE

In any loan transaction the bank undertakes to pay the loan amount over to the borrower at

a deﬁned point in time. In return the borrower undertakes to pay the bank loan interest and

repayments on the dates agreed. At ﬁrst therefore there is a cash ﬂow from the bank to the

10 Risk-adjusted Lending Conditions

borrower, and then one or several cash ﬂows from the borrower to the bank. Complicated loan

arrangements in respect of amounts paid out and amounts paid back may be broken down into

their individual components as such.

Because of the shortfall risk, however, the cash ﬂows nominally agreed in the loan contract

do not ﬂow to the bank, but only their probability values which lie, in terms of value ﬁgures,

between zero and their face value.

Following Volkart’s expositions on the assessment of ﬁnance contracts [VOLK93, S. 340ff],

the sum of the present values of the breakdown values of the cash ﬂows under the loan contract

have to be determined, in order to be able to assess the loan that has been paid over. We assume

here that the risk rate is correctly calculated, if this total corresponds to the loan paid over. As

the correctly calculated breakdown values of the cash ﬂows, which are already an expression

of the correctly calculated shortfall risk, may be regarded as implicitly more secure payments,

they must be discounted by the corresponding risk-free standard interest rate.

Expressed mathematically, the model for calculating the risk surcharge will now run as

follows:

L =

n

¸

j =1

ψ

j

· C

j

(1 +i

s

)

j

(1.2)

L = paid out loan amount

n = number of periods of loan maturity

ψ

j

= probability of cash ﬂow C

j

C

j

= cash ﬂow face value after period j

i

s

= risk-free standard interest rate

Here the consideration that it is not just the face values, but the breakdown values of the

cash ﬂows that are discounted, is fundamental. The variable ψ

j

acquires decisive signiﬁcance.

We will go into this more closely in Chapter 3.

As demonstrated below, the price of the risk-encumbered loan L is expressed in the nominal

values of the cash ﬂows C

j

.

Continuing, we assume that there is a uniform risk-free standard rate of interest i

s

as a

reference rate for each loan transaction, which is calculated as follows:

i

s

= f + p (1.3)

The reason for the standard rate of interest i

s

is the following: at a purely theoretical level

a borrower might think that it presents no shortfall risk at all for the bank when any loan is

granted. There is thus absolutely certainty that the debt servicing will be performed according

to contract. The bank only has to charge such a debtor the reﬁnancing costs and an appropriate

proﬁt contribution. The rate of shortfall risk is precisely zero. In this exposition the standard

rate of interest will therefore take over the function of the risk-free rate of interest in ﬁnancial

market theory, without being identical to it. As will be shown in Section 7.3, this action is

permissible as a borrower’s shortfall risk, under the theory explained here, is independent of

the standard interest rate and of the risk-free interest rate, respectively.

The standard rate of interest does not have to be identical for all borrowers, although it

may remain the same for the term. As already mentioned footnote 1, the statutory underpin-

ning by own resources may be varied according to ﬁnancial standing, providing the statutory

Introduction 11

requirements are, on average, met. The proﬁt contribution rate does not have to be identi-

cal either. Customers of better ﬁnancial standing usually give rise to less loan assessment

expenditure and vice versa, which justiﬁes making such a distinction overall.

Following Brealey and Myers [BRMY96, S. 35] the amount to the right of the equals sign

in equation (1.2) may be considered to be the present value adjusted for risk in respect of

ﬁnancial standing, and the credit paid out to the left of the equals sign in equation (1.2) may be

considered to be the investment the bank has to make in the loan transaction concerned. The

net present value of the loan transaction is therefore [BRMY96, S. 13, S. 35]:

NPV

Credit

=

n

¸

j =1

ψ

j

· C

j

(1 +i

s

)

j

− L (1.4)

Using equation (1.4) the requirement for a loan transaction consistent with risk may be for-

mulated as follows: a loan transaction is reckoned to be entirely consistent with risk if its

net present value calculated at the standard rate of interest is equal to zero, at which it is the

expectation values of the cash ﬂows, not their nominal values, that are discounted.

As the risk-free standard rate of interest in equation (1.3) already contains a proﬁt contribu-

tion rate for the bank extending the credit, the appropriate earnings for the bank are therefore

built into the calculation.

1.7 ASSUMPTIONS

At this point the assumptions made in this study are listed again:

1. Only the shortfall risk as deﬁned by Kilgus [KILG94, S. 69] is calculated.

2. The price of a loan transaction may be expressed in one integrated loan interest rate that

incorporates all the elements of the price.

3. The bank’s ﬁnancing costs and the costs of a loan transaction in relation to market and

liquidity risks accrue in proportion to the amount of credit that is taken up, and can therefore

be imputed in an integrated ﬁnancing cost rate for the loan transaction concerned.

4. The costs of processing a loan, the behavioural and operational risks do not arise in propor-

tion to the extent to which the credit is taken up, and have to be imputed into the revenue

from, and be covered by, the highest possible proﬁt contribution rate. The revenue from the

loan transaction concerned will be calculated via a calculation of proﬁt contribution rate,

if applicable calculated at a number of different levels or stages, starting from the proﬁt

contribution normally achieved.

5. The shortfall risk hedging rate is dependent on the borrower’s ﬁnancial standing.

6. The shortfall risk hedging rate of a loan will have been correctly pitched, if the total of

the expectation values of the cash ﬂows, discounted by the standard rate of interest, are

precisely equal to the loan amount extended under the loan agreement.

7. For each loan transaction there is an integrated standard rate of interest as a reference rate.

Assumption 1 is a voluntary restraint for the purposes of this study. It is, however, not thereby

implied that no other risks exist (for example, market, liquidity, behavioural and operational

risks). The only assumption here is that these risks have already been taken into account (see

below).

12 Risk-adjusted Lending Conditions

Assumption 2 is in line with normal banking practice up until now. In the last few years,

however, some banks have been attempting to subdivide this principle by introducing fees

for administering loans and investigating credit status. Assumption 2 nonetheless implies no

qualiﬁcation to the generality that such fees have traditionally only been applied, as precon-

ditions of the loan, at the time the loan agreement is concluded or the loan is made available.

Assumption 2 is valid for the duration of the loan.

Assumptions 3 and 4 are in line with the current practice of any bank in terms of business

management. We will not go into the related difﬁculties in application here, in view of the

reservations outlined in Section 1.2.

Assumption 5 is self-evident, and forms the main subject of this study.

Assumption 6 is the fundamental assumption of the model.

Assumption 7 is not made to the absolutely fullest extent for each loan transaction. According

to how many lines of credit exist between the borrower and the bank, and to their various

time-scales, this assumption may not be completely sustainable. It will nevertheless be made

frequently in comparable situations, in order to simplify our considerations (cf. for example

[BRMY96, S. 36]. Complicated termstructures of standard interest and/or discount rates under

equation (1.2) may also be converted to one integrated rate, at least approximately, in the course

of calculating the rate of yield to maturity (cf. [BRMY96, S. 646–649]).

1.8 TESTING THE MODEL

A mathematical model is only as good as the extent to which it is able to reﬂect real life, and

this study is no exception. The forecasts made using it have to be capable of being checked

against examples of losses on loans that actually occur, in order for it to be possible to make

reliable use of this model.

The Basel Committee for bank supervision has tested existing models for measuring lending

risks, and has come to the conclusion that there cannot yet be any question of applying such

models in the sphere of supervision. Two problem areas in particular led to this conclusion.

Data limitations: banks and researchers alike report data limitations to be a key impediment to

the design and implementation of credit risk models. Most credit instruments are not marked to

market, and the predictive nature of a credit risk model does not derive froma statistical projection

of future prices based on a comprehensive record of historical prices. The scarcity of the data

required to estimate credit risk models also stems from the infrequent nature of default events and

the longer-termtime horizons used in measuring credit risk. Hence in specifying model parameters,

credit risk models require the use of simplifying assumptions and proxy data. The relative size

of the banking book —and the potential repercussions on bank solvency if modelled credit risk

estimates are inaccurate —underscore the need for a better understanding of a model’s sensitivity

to structural assumptions and parameter estimates. [BCBS99, S. 1]

Model validation: the validation of credit risk models is fundamentally more difﬁcult than the

back-testing of market risk models. Where market risk models typically employ a horizon of a few

days, credit risk models generally rely on a time frame of one year or more. The longer holding

period, coupled with the higher conﬁdence intervals used in credit risk models, presents problems

to model-builders in assessing the accuracy of their models. By the same token, a quantitative

validation standard similar to that in the Market Risk Amendment would require an impractical

number of years of data, spanning multiple credit cycles. [BCBS99, S. 2]

Introduction 13

In addition the Basel Committee set out what precautions should be borne in mind when

testing models.

The components of model validation can be grouped into four broad categories:

(a) back-testing, or verifying that the ex-ante estimation of expected and unexpected losses is

consistent with ex-post experience;

(b) stress testing, or analysing the results of model output given various economic scenarios;

(c) assessing the sensitivity of credit risk estimates to underlying parameters and assumptions;

and

(d) ensuring the existence of independent review and oversight of a model.

At present, few banks possess processes that both span the range of validation efforts listed and

address all elements of model uncertainty. This suggests that the area of validation will prove to

be a key challenge for banking institutions in the foreseeable future. [BCBS99, S. 50]

In Chapter 12 we will explain what precautions should be borne in mind in order to meet the

problems outlined above when applying the model presented here. In this respect it is timely

to stress, at this juncture, that methods for testing mathematical models measuring credit risks

have still to be developed.

At present, there is no commonly accepted framework for periodically verifying the accuracy of

credit risk models; going forward, methods such as sensitivity testing are likely to play an important

role in this process. Finally, it is important to note that the internal environment in which a model

operates —including the amount of management oversight, the quality of internal controls, the

rigour of stress testing, the reporting process and other traditional features of the credit culture —

will also continue to play a key part in the evaluation of a bank’s risk management framework.

[BCBS99, S. 6]

The introduction and above all the regular examination of a mathematical model for mea-

suring credit risks represents a major challenge for any bank, and is at present still closely

linked to development of principles. In this respect it is indeed timely to emphasise that the

method introduced here may only be examined for its validity in the future, if a bank is indeed

prepared to create the necessary preconditions for doing so now.

The Basel Committee made it its primary concern to examine whether there are already

models suitable for the purposes of supervision as required by law. This may, however, be

more of a second step. Some bank has ﬁrst to succeed in furnishing proof that it is in a position

reliably to measure, and in turn to forecast, its credit risks with a mathematical model. Only

then may the extent to which such a model might also be suitable for use for the purposes of

supervision, as required by law, be examined.

1.9 LOAN EXPOSURE MODELS

To conclude this introduction, the model developed in this book is set in the context of loan

exposure models existing hitherto. To do this we must ﬁrst brieﬂy review the methods that

have been used to date.

The previous models for deﬁning the shortfall risk of a loan can be classiﬁed into two groups:

r

Classical methods prior to the development of ﬁnancial theory.

r

Methods based on modern ﬁnancial theory that has been developed since the 1970s.

14 Risk-adjusted Lending Conditions

The previous courses of action —divided into the two groups mentioned —are presented in

outline in the following two subsections. The model described in this book is put into context, in

relation to the earlier models, in Subsection 1.9.3. At this juncture it is not, however, possible to

go all that much into detail, and we accordingly refer the reader to the bibliography. For instance

Cossin and Pirotte [COPI01] give a good overall view of credit risk models in existence. The

expositions that follow in 1.9.1 and 1.9.2 are based substantially on their compilation.

1.9.1 Classical Methods (see [COPI01, p. 91])

Most of the classical literature on credit risk tends to bear on traditional actuarial methods of

credit risk (see [CAOU98] for a survey of these; see also for a critical approach [DUFF95a/b]).

Although these methods are widely used in banks, they present some difﬁculties. The basic

principle of this type of approach is to estimate (often independently) the value of the contract

at possible default times.

Rating agencies are standard sources for default probabilities. Techniques used to forecast

default probabilities for individual ﬁrms are described in [ALTM77]. Methodologies have

evolved from the calculations of mortality rates to the calculation of rating category migration

probabilities. These probabilities (usually organised in so-called transition matrices) consist

of the probabilities of downgrade and upgrade by rating category. These calculations are now

frequently used by professionals.

As stressed by [DUFF95a], end users tend to develop Monte Carlo simulations without

taking into account the uncertainties in the models used to generate the estimates. Second,

they rarely take into account the correlations among probabilities of default and estimates of

possible losses. These correlations certainly affect the results. One can expect, for example,

exposures linked to derivatives to rise with the volatility of the markets. But it is also at such a

time that probabilities of default will arise. Unfortunately, historical correlations are difﬁcult

to obtain empirically. Some try to overcome this difﬁculty by using advanced analysis methods

such as neural networks (see, for example [TRTU96]).

Nonetheless, all these methods face the major difﬁculty of being strongly dependent on

historical estimates of credit risk dynamics. They are still a useful basis of information to start

from, but ﬁnancial theory has now provided us with more powerful analytical tools.

1.9.2 Modern Credit Risk Analysis Based on Financial Theory

(see [COPI01, pp. 9–13])

Modern credit risk analysis, on the other hand, is along the line of the continuous develop-

ment of ﬁnancial research on the integration of uncertainty. Broadly speaking, the investor

faces risks that are categorised as market risks, credit risk, country risk, and operational risk.

Modern appraisal of credit risk follows directly from the advances that have been made for the

management of market risks. To understand why the latter has been such a preoccupation in

modern ﬁnance, let us introduce some chronology about market risks, their development, and

the needs that have increased with respect to them.

Market risks integrate interest rate risk, exchange rate risk and stockmarket risk. Interest

in market risks began ﬁrst with the development of stock exchanges and banking systems in

Introduction 15

most of the developed countries. With the end of the Bretton Wood agreements, exchange

rates were then allowed to ﬂoat causing volatility in interest rates. From the late 1970s, many

economic studies were undertaken giving rise to what has been called since, ﬁnancial theory.

On the practitioners’ side, many forms of contracts were proposed to the investors to mitigate

the increasing volatility on the market, with special clauses allowing them to be optionally

protected against changes in the term structure. This produced contracts that not only were

sensitive to changes in market factors but also showed discontinuities in them. Therefore,

the classical present value of coupon payments and the simple calculation of durations and

convexities appeared to be insufﬁcient to monitor and manage those risks.

The interest of academics in developing new theories to modelise the uncertainty of market-

risk phenomena has led to a sophisticated set of ﬁnancial tools inspired from mathematics

and physics. The evolution happened on two grounds: ﬁnancial theory, mostly driven from

economics theory at that point; and the inclusion of sophisticated mathematics.

On the ﬁnancial theory side, most of the research attempted to give a value to the market

price of risk or market risk premia. ‘Market’ because the risk comes from market variables

and also because, in order to ﬁnd a unique market risk premium for each factor, the general

hypothesis being made is that only systematic risk (the undiversiﬁable one) is priced.

On the mathematical side, much of today’s inheritance comes from the early introduction

of stochastic calculus (well known in physics for its application to problems such as health

propagation) into modern ﬁnance. The contribution of stochastic calculus is ﬁrstly its capacity

to produce a deterministic solution out of an uncertainty that is modelled as a random process.

The dynamics of the unexpected part of the uncertainty is not deterministically speciﬁed from

the beginning, as is the case with chaos theory. Moreover, stochastic calculus allows the reﬁning

of the time space into inﬁnitesimal pionts as a limit of the discrete-time approach. Let us take

an example. Suppose that we want to draw the evolution of the stock price and the terminal

values that it can take in one month. In discrete time, we have to choose the number of time

steps up to the maturity of one month, while in continuous time there is an inﬁnity of time steps

guiding the stock price to its terminal value. In the latter case, there are no discontinuities at all

in the evolution of the stock price. In continuous time, the process does not execute jumps to

two adjacents values but rather changes in a very small period of time to very small different

values.

The continuous-time framework is very useful because it enables much more easily closed-

form solutions to speciﬁc ﬁnancial problems to be obtained, while the discrete approach is

still of great help to visualise the choices to make through time. But these choices are made

on speciﬁc dates while they are made continuously in continuous time. As noted in Merton’s

articles, two basic assumptions have to be made to justify the use of continuous-time approaches

in the portfolio selection problems of modern ﬁnance:

Assumption 1: Capital markets are open at all times meaning that agents can trade continu-

ously.

Assumption 2: The stochastic processes generating the state variables can be described by

diffusion processes with continuous sample paths.

The contribution of Merton resides in his capacity, at that time, to relate ﬁnancial theory and

the continuous-time approach introducing the well-known continuous-time ﬁnance. Before its

16 Risk-adjusted Lending Conditions

emergence, ﬁnancial theory was limited to static theories. Continuous-time ﬁnance allowed the

restatement of previous problems dynamically, showing how theories such as the CAPM are

inﬂuenced if the investor is now allowed to behave dynamically. The investor is now allowed

to react continuously to changes in the environment rebalancing and hedging his positions

through time. This ability should be taken into account along with transaction costs to show

how strategies can be optimised from the beginning. Moreover, having this possibility means

that the investor has a non-executed option on future allocation, which has a price therefore at

time 0.

In 1973, Black and Scholes were able to price such complex products as standard call and put

options. The contribution of Merton is substantial and visionary, giving rise in the 1980s and

1990s to a rush into the design of derivatives products of increasing complexity. All these tools

enable us today to price securities subject to these risks and to design sophisticated contingent

claims on the same securities.

Now that market risks seems to be well encompassed, research has turned to credit risk.

This interest also ﬁlls a need. The wave of developments for market risks has engendered a

sudden awareness in other ﬁelds, precisely about credit risk, for several reasons. One of them

which is very relevant is the fact that differences between European currencies are vanishing

with the appearance and global use of the Euro in ﬁnancial markets. For European currencies’

denominated bonds, credit risk then becomes the main determinant of the spread of a corporate

bond yield over the risk-free rate. Here, since a linkage is directly being made between the

interest rate risk and the credit risk, we cannot stand on traditional actuarial approaches for the

credit risk part to price those bonds. Another reason is the huge movements in credit standing

characterising the end of 1998. From 1996, we observe a continuously growing number of

defaults occuring. With the globalisation wave, falling economies’ effects propagate strongly

and quickly to high-grade economies. Investors are thus far aware that credit risk is a real

problemand that it cannot be measured and monitored on a standalone basis. Therefore, market

efﬁciencydemands that the potentialityof credit losses must be accuratelyestimatedandpriced.

1.9.3 The Model Presented Here Seen in Relation to Previous Models

The model presented here has two bases. On the one hand it is based on the groundwork

developed in Part II of this book on the connections between the shortfall risk of the borrower,

the credit shortfall risk, the breakdown distribution rate, the standard rate of interest without

taking the credit shortfall risk into account, and the risk-adjusted loan interest rate. These

connections are derived with the aid of classical probability calculus. On the other hand it is

based on the original Black/Scholes model [BLSC73] for the evaluation of company debts. The

combination of these two bases leads to an algorithm that allows the risk of each individual

debt position in any company balance sheet to be assessed on the basis of ﬁnancial theory

(or, more precisely, on the basis of option price theory), in line with shortfall risk. This is a

substantial advance on the Black and Scholes account [BLSC73]. That an extension of their

model is able to achieve such results does, however, also demonstrate how revolutionary their

reﬂections already were.

This algorithm results in one receiving, for each debt position in the company balance

sheet, not only the borrower’s shortfall risk but also the credit shortfall risk. From these

Introduction 17

one can calculate, for each position, the probability value —according to the model —of the

breakdown distribution rate. Using the standard rate of interest, we arrive at the risk-adjusted

loan interest rate.

The great advantage of these detailed results lies in that the fact that they may now be

combined with classical ways of looking at things. Let us assume now that the breakdown

distribution rate calculated according to the model does not, for instance, agree with previous

experience in the case of a concrete loan transaction —or that it has still to be adjusted (cf., for

example, Section 7.9). The borrower’s shortfall rate determined according to the model may

now be combined with the modiﬁed breakdown distribution rate, in order to determine the

credit shortfall rate more precisely in the case concerned. In so far as it may be necessary to

do so, our model thus allows for situations to be examined using a combination of classical

methods and methods based on ﬁnancial theory. When linked in with essential professional

experience in lending business operations, it is thus possible to calculate the best possible

estimations of loan exposure for given loans (in so far as this is, ex ante, possible at all).

The method put forward here is thus indeed based on ﬁnancial theory, with its associated

advantages. Owing to its ﬂexibility, however, it also allows for combination with the results

of classical approaches, where this is appropriate and meaningful. In this way the experienced

professional obtains an effective instrument for assessing and evaluating loans.

2

Rating System

The necessity for a rating system is explained in Section 2.1, the shortfall risk is deﬁned

numerically in Section 2.2, and the credit-worthiness key ﬁgure is deﬁned in Section 2.3.

A numerical rating system is introduced in Section 2.4 and elaborated in Section 2.5.

2.1 THE NEED FOR A RATING SYSTEM

Kilgus himself emphasises the need to subdivide a bank’s borrowers into various risk categories

(ratings) [KILG94, S. 70]. We will explain below why this applies here too, and what has to

be done to deﬁne a borrower’s rating mathematically.

The fundamental difﬁculty in deﬁning a borrower’s future shortfall risk consists in putting

forward an ex-ante forecast on the basis of ex-post facts. Historical facts do provide useful

indications for this, but the future cannot be considered simply by extrapolating the past. There

are, moreover, borrowers that may not yet have any kind of track record. This brief reﬂection

indicates right away that a borrower’s future shortfall risk cannot be deﬁned as precisely

as one would like, and that a soundly deﬁned estimated value will always be required to a

greater or less degree. Anything else would be crystal-ball gazing, as reproducible experiments,

such as are possible in the natural sciences, are beyond the scope of the science of business

management.

The introduction of a rating system offers a way out of this dilemma (cf. also [CART98]).

Borrowers are subdivided into groups with comparable shortfall risks. What this entails pre-

cisely is explained in Section 2.4. The individually calculated shortfall risk is not used for

calculating the loan price, but the maximum value according to the rating level. The same

imputed shortfall risk is therefore allocated to all borrowers with the same rating. This action

is based on the realisation outlined above that future values can only be estimated, and that any

other action would simply mean faking apparent exactitudes. The use of the maximum value

for a group is consistent with the principle of conservatism.

The reﬁnement of the rating system chosen depends essentially on how much cost it is

intended to incur on the estimated accuracy of future shortfall risks. This in turn depends on

how competitive a bank seeks to be in the market for any particular loan product. The market’s

sensitivity to price has to be reﬂected in the rating system. On the other side of the coin, more

expenditure on estimation means higher processing costs and with that of necessity a larger

proﬁt contribution rate p in the calculation of the price of the loan product.

The question of what might be the optimum estimated accuracy versus the estimated expen-

diture needed to obtain it can only be answered by staying permanently close to what is going

on in the marketplace. There is simply no easy answer to this.

20 Risk-adjusted Lending Conditions

2.2 DEFINING SHORTFALL RISK IN TERMS OF FIGURES

It is assumed here that a shortfall risk ρ in terms of ﬁgures can be assigned to each borrower. ρ

is therefore deﬁned as the probability that a borrower will, within some future period of time,

no longer be in a position fully to meet its commitments to a bank. Unless otherwise speciﬁed,

we will always be considering a time period of one year. ρ may therefore be assumed to have

values within:

0 ≤ ρ ≤ 1 (2.1)

ρ = Shortfall risk

Onthe other side of the coin, the chance χ maybe deﬁnedas the chance of anyborrower being

able to meet its commitments at any time within any future period of time. χ is thus deﬁned as

the probability of any borrower being in a position, at any time within any future period of time,

of meeting its commitments to the bank in full. Unless otherwise speciﬁed, we will always be

considering, a time period of one year. χ is therefore assumed to have values between:

0 ≤ χ ≤ 1 (2.2)

χ = Survival chance

It is worth noting that ρ and χ are not identical with Ψ

j

on the strength of these deﬁnitions

(see Section 1.6). The correlation between these three values is derived later in this chapter.

Any borrower may be in a position, within the same period of time, either to meet its

commitments to the bank in full, or not be in such a position: there is no third option. For any

borrower and period of time the following deﬁnition therefore holds good:

ρ +χ = 1 (2.3)

Bankruptcycases are far frommeaningthat all the moneylent has beenlost tothe bank. It may

rather be that an expectation of a percentage recovery B may enter into the equation concerned.

The following correlation applies, in which i represents the interest that has accumulated:

b =

B

L · (1 +i )

(2.4)

B = breakdown distribution probability value

b = breakdown distribution rate probability

The breakdown distribution rate is in turn a probability dimension. It is therefore worth:

0 ≤ b ≤ 1 (2.5)

The expectation value of loss in cases of bankruptcy amounts therefore to (1 −b) L.

Multiplied by the probability of bankruptcy occurring ρ, the credit shortfall risk ρ

∗

is as follows:

ρ

∗

= (1 −b) · ρ (2.6)

0 ≤ ρ

∗

≤ 1 (2.7)

ρ

∗

= Credit shortfall risk

Rating System 21

Following on fromequation (2.3) the value of χ

∗

is deﬁned, purely arithmetically, as follows:

χ

∗

= 1 −ρ

∗

(2.8)

0 ≤ χ

∗

≤ 1 (2.9)

χ

∗

= survival chance regarding the breakdown distribution rate

2.3 DEFINING THE CREDIT-WORTHINESS KEY FIGURE

Under this deﬁnition the credit shortfall risk, and with it the ﬁnancial standing of a borrower,

can be expressed in ﬁgures by ρ

∗

, where the value of ρ

∗

lies between zero and one. As it is

normally unwieldy to make a presentation of orders of magnitude using decimal fractions, we

make the additional deﬁnition here of the credit-worthiness key ﬁgure κ as the inverse of ρ

∗

:

κ =

1

ρ

∗

(2.10)

The following range applies for κ on the basis of the range for ρ

∗

:

1 ≤ κ ≤ ∞ (2.11)

2.4 EXAMPLE OF A RATING SYSTEM IN TERMS OF FIGURES

The recognised rating agencies subdivide the issuers of loans in such a way that a loan issuer

is in principle no different from a borrower. The rating categories are in this sense deﬁned

qualitatively as, for example, in the case of Moody’s [MOOD90, S. 14/15].

In contrast to this, we are concerned here to capture the shortfall risk of a borrower in

terms of a numerical probability. The rating categories in this example are therefore deﬁned

numerically. The possible values of 0 ≤ ρ

∗

≤ 1 are thus assigned to rating categories.

For the purposes of illustration, such a rating system is developed at this point, in line with

the following principles (all sorts of other principles could of course be imagined!):

1. 12 levels (AAA, AA, A, BBB, BB, B, CCC, CC, C, DDD, DD, D).

2. As good borrowers in terms of credit-worthiness are substantially more price sensitive than

bad ones, level AAA must be selected more narrowly than level AA, and so on: a factor of

2 is used at this point, i.e. level AA is twice as wide as level AAA, and so on.

Supported by the above principles, the following relative widths ensue for the individual

levels:

AAA = 1 BBB = 8 CCC = 64 DDD = 512

AA = 2 BB = 16 CC = 128 DD = 1024

A = 4 B = 32 C = 256 D = 2048

The total of these relative widths comes to:

11

¸

j =0

2

j

= 4095 (2.12)

22 Risk-adjusted Lending Conditions

The following effective level widths ensue:

AAA = 1/4095 = 0.0244%

AA = 2/4095 = 0.0488%

A = 4/4095 = 0.0977%

BBB = 0.1954% CCC = 1.5629% DDD = 12.5031%

BB = 0.3907% CC = 3.1258% DD = 25.0061%

B = 0.7814% C = 6.2515% D = 50.0122%

and the following applies:

11

¸

j =0

2

j

4095

= 1 (2.13)

This now results in the numerical rating system shown in Table 2.1.

A simpliﬁed rating system for less competitive markets may be drawn up, derived from

Table 2.1, with only four levels as shown in Table 2.2.

Table 2.1 Rating system used in the context of this book

Value of ρ

∗

(%)

Rating from to

ρ

∗

according to

rating level (%)

κ according to

rating level (rounded)

AAA 0.0000 0.0244 0.0244 4095

AA 0.0244 0.0733 0.0733 1365

A 0.0733 0.1709 0.1709 585

BBB 0.1709 0.3663 0.3663 273

BB 0.3663 0.7570 0.7570 132

B 0.7570 1.5385 1.5385 65

CCC 1.5385 3.1013 3.1013 32

CC 3.1013 6.2271 6.2271 16

C 6.2271 12.4786 12.4786 8

DDD 12.4786 24.9817 24.9817 4

DD 24.9817 49.9878 49.9878 2

D 49.9878 100 100 1

Table 2.2 Simpliﬁed rating system

Value of ρ

∗

(%)

Rating from to

ρ

∗

according to

rating level (%)

κ according to

rating level (rounded)

A 0.0000 0.1709 0.1709 585

B 0.1709 1.5585 1.5585 65

C 1.5585 6.2271 6.2271 8

D 6.2271 100 100 1

Rating System 23

The relative widths of the individual rating levels here come to A = 1, B = 8, C = 64 and

D = 512. The factor from level to level thus amounts to 2

3

= 8, as three times fewer levels

occur than in the preceding example.

2.5 AMPLIFICATION OF THE RATING SYSTEM FOR VERY

COMPETITIVE MARKETS

The markets for loans todebtors withveryhighﬁnancial standing, above all, are oftenextremely

competitive. The rating system presented in Table 2.1 might therefore be insufﬁciently precise

for such markets. It should therefore be pointed out that the system presented in Table 2.1 may

be further reﬁned. There are three times fewer levels in the simpliﬁed system in Table 2.2 than

in the systemin Table 2.1. The relative width of the individual levels therefore grows by a factor

of 8 = 2

3

; the exponent 3 being attributable to the number of levels being three times smaller.

Analogously, this factor comes to 2

1/3

, in the case of a system with three times as many levels,

being the third root of 2. By analogy with the method of calculation in the preceding section,

one thus obtains the reﬁned rating system shown in Table 2.3 when the number of levels is

tripled.

Table 2.3 could be developed over all levels to D. This would, however, be superﬂuous,

as the markets for loans to borrowers of lower ﬁnancial standing become progressively less

competitive.

It must be noted that ρ

∗

and κ are in each case identical for the AAA, AA, A and BBB

ratings in Table 2.2, and for the AAA-, AA-, A- and BBB- ratings in Table 2.3.

Table 2.3 Reﬁned rating system

Value of ρ

∗

(%)

Rating from to

ρ

∗

according to

rating level (%)

κ according to

rating level (rounded)

AAA+ 0.0000 0.0063 0.0063 15755

AAA

∗

0.0063 0.0143 0.0143 6971

AAA− 0.0143 0.0244 0.0244 4095

AA+ 0.0244 0.0371 0.0371 2694

AA

∗

0.0371 0.0531 0.0531 1883

AA− 0.0531 0.0733 0.0733 1365

A+ 0.0733 0.0986 0.0986 1014

A

∗

0.0986 0.1306 0.1306 765

A− 0.1306 0.1709 0.1709 585

BBB+ 0.1709 0.2217 0.2217 451

BBB

∗

0.2217 0.2857 0.2857 350

BBB− 0.2857 0.3663 0.3663 273

Part II

Mathematical Foundations of the Model Mathematical Foundations of the Model

Probability model: Development of ψ

j

Calculation of the shortfall risk hedging rate in the special case

of shortfall risks being constant

Calculation of the shortfall risk hedging rate in the general case

of variable shortfall risk

Shortfall risk on uncovered loans on the basis of statistics

3

Probability Model: Development of ψ

j

As may be inferred from the basic equation (1.2), determining the probability ψ

j

of cash ﬂow

C

j

being fulﬁlled is of decisive importance for the model we are describing. The correlation

between the shortfall risk ρ and the survival chance χ and of the probability of fulﬁlment ψ

j

will be derived in Section 3.1, with the aid of probability calculus.

In Section 3.2 we will show how the shortfall risk and survival chance might be converted

over various terms. Conclusions may be drawn from the results of Section 3.1 for loans that

are unlimited in time and for ‘reasonable’ terms in relation to the shortfall risk ρ. This will

be presented in Section 3.3. For the sake of clarity the results of Chapter 3 will be presented

again in Section 3.4.

3.1 DETERMINING THE PROBABILITY OF CASH FLOWS

BEING FULFILLED

What we are concerned to do below is develop a model for determining probabilities using

equation (1.2). The components needed for this model are deﬁned as follows (with the verb ‘to

default’ being used as a synonym for the sentence ‘to no longer be able to meet commitments

to the bank in full’):

r

n is the number of periods in the term of the loan.

r

j represents the period concerned: 1 ≤ j ≤ n.

r

ρ

j

is the probability of the borrower defaulting within period j .

r

χ

j

is the probability of the borrower not defaulting within period j .

r

ϕ

j

is the probability of the borrower defaulting between the ﬁrst period and period j .

r

ε

j

is the probability of the borrower not defaulting between the ﬁrst period and period j .

r

ϕ(n) is the probability of the borrower defaulting at some point during the term of the loan

of n periods.

r

ε(n) is the probability of the borrower not defaulting during the whole of the term of the

loan of n periods.

The following correlations apply by deﬁnition (see also equation (2.3)) (cf. [BOHL92,

S. 312]):

ρ

j

+χ

j

= 1 (3.1)

ϕ(n) +ε(n) = 1 (3.2)

The results are derived in Appendix 4. They are:

ψ

j

= ε

j

=

j

¸

k=1

χ

k

=

j

¸

k=1

(1 −ρ

k

) (3.18)

28 Risk-adjusted Lending Conditions

In the special case of shortfall risks being constant, the following applies:

ψ

j

= χ

j

= (1 −ρ)

j

if χ

1

= · · · = χ

n

= χ (3.19)

3.2 MATURITY TRANSFORMATION

What we are concerned to prove now is that the values of the probabilities ρ and χ under the

above deﬁnitions, in relation to the same borrower, are dependent on the length of the period.

We will therefore in this section distinguish between the values for ρ and χ in relation to one

month (ρ

m

, χ

m

), one quarter (ρ

q

, χ

q

), a period of six months (ρ

s

, χ

s

), and to one year (ρ

y

, χ

y

).

Let us assume a loan with a term of 12 months. Then the following applies in general,

according to equation (3.17) (see Appendix 4):

ϕ(12M) = 1 −

12

¸

j =1

χ

mj

= 1 −

12

¸

j =1

(1 −ρ

mj

) (3.20)

and in the special case using equation (3.13) (see Appendix 4):

ϕ(12M) = 1 −χ

12

m

= 1 −(1 −ρ

m

)

12

if ρ

m1

= · · · = ρ

m12

= ρ

m

(3.21)

By analogy, the following applies, according to equation (3.14) (see Appendix 4):

ε(12M) =

12

¸

j =1

χ

mj

(3.22)

and according to equation (3.15) (see Appendix 4) in special cases:

ε(12M) = χ

12

m

if χ

m1

= · · · = χ

m12

= χ

m

(3.23)

On the basis of the deﬁnition, however, ϕ(12M) is the same as ϕ

y

, and ε(12M) is the same

as ε

y

.

We can thus now detail the following maturity transformations by using the analogy for the

special case ρ

m1

= · · · = ρ

m12

= ρ

m

and χ

m1

= · · · = χ

m12

= χ

m

. . . . .

ρ

y

= 1 −(1 −ρ

s

)

2

= 1 −(1 −ρ

q

)

4

= 1 −(1 −ρ

m

)

12

(3.24)

ρ

s

= 1 −

1 −ρ

y

= 1 −(1 −ρ

q

)

2

= 1 −(1 −ρ

m

)

6

(3.25)

ρ

q

= 1 −

4

1 −ρ

y

= 1 −

1 −ρ

s

= 1 −(1 −ρ

m

)

3

(3.26)

ρ

m

= 1 −

12

1 −ρ

y

= 1 −

6

1 −ρ

s

= 1 −

3

1 −ρ

q

(3.27)

χ

y

= χ

2

s

= χ

4

q

= χ

12

m

(3.28)

χ

s

=

√

χ

y

= χ

2

q

= χ

6

m

(3.29)

χ

q

=

4

√

χ

y

=

√

χ

s

= χ

3

m

(3.30)

χ

m

=

12

√

χ

y

=

6

√

χ

s

=

3

√

χ

q

(3.31)

Probability Model 29

Further maturity transformations may be worked out analogously, under this procedure.

3.3 CONCLUSIONS

Using the special case ρ

1

= · · · = ρ

n

= ρ as an example, some correlations may be shown

that ensue as a direct consequence of the properties of geometric series.

3.3.1 The Case of a Loan being Granted Indeﬁnitely

Under Swiss banking practice, current account credit in particular often has no time limit set on

it contractually. Equation (3.12) is reformulated as follows for an inﬁnite number of periods:

ϕ(∞) =

∞

¸

j =1

(1 −ρ)

( j −1)

· ρ = ρ ·

∞

¸

j =1

(1 −ρ)

( j −1)

ϕ(∞) = ρ ·

1

1 −(1 −ρ)

= ρ ·

1

ρ

= 1 if ρ > 0! (3.32)

Equation (3.32) does, however, only apply for ρ > 0, because a zero/zero division otherwise

arises. The following applies under equation (3.13) (see Appendix 4) for the case where ρ = 0:

ϕ(∞) = 1 −(1 −0)

∞

= 1 −1

∞

= 0 if ρ = 0 (3.33)

From this one may draw the conclusion that every borrower defaults at some point if the

time span for which credit is granted has an inﬁnite number of periods; otherwise its shortfall

risk ρ would be precisely zero! From this it follows that every loan must be limited in time,

as borrowers with shortfall risks of zero do not exist! In the case of overdraft facilities, this

requirement becomes relative, however, as under normal contract clauses notice to terminate

the facility may be given at any time. In the course of operational banking this leads to the need

to review every loan, from time to time, according to its shortfall risks. If there are changes

in the risk position, then appropriate action must be taken: adjustment of the rate of interest,

change in the frequency of review, or possible notice.

3.3.2 Reﬂections on the Success Chance ε(n)

In Figure 3.1 ε(n) is given for various values of ρ, a logarithmic scale having been selected for

the n axis. n = 1/ρ is a characteristic number of periods in this for each curve. It follows from

the illustration that all ε(1/ρ) for ρ →0 have about the same value. The maximumε(1/ρ) for

ρ →0 may be calculated using equation (3.15) as an example (see Appendix 4).

lim

ρ→0

ε(1/ρ) = lim

ρ→0

(1 −ρ)

1/ρ

(3.34)

30 Risk-adjusted Lending Conditions

1 10 100 1000 10000

100%

50%

0%

100%

50%

0%

(n) (n = 1/nr)

36.8%

60.7%

90.5%

ρ = 10% ρ = 1% ρ = 0,1%

n = 10

n = 2

n = 1

n

Figure 3.1

The substitution x = 1/ρ leads to [DMK/DPK92, S. 33]

lim

ρ→0

ε(1/ρ) = lim

x→∞

ε(x) = lim

x→∞

1 −

1

x

x

=

1

e

1

e

≈ 0.368 = 36.8% (3.35)

which means that a loan with term of 1/ρ periods has a maximum probability of success of

about 36.8%, or conversely, a minimum probability of loss of about 63.2%. It looks better

if shorter periods of time related to n = 1/ρ are considered. By introducing the ratio ν, the

following may be set out, analogously to the above, for n = 1/(ν · ρ)

lim

ρ→0

1

ν · ρ

= lim

ρ→0

(1 −ρ)

1

ν·ρ

= lim

x→∞

1 −

1

ν · x

x

=

1

e

ν

(3.36)

This leads for various values of ν to the following values for ε(1/(ν · ρ))

max

and

ϕ(1/(ν · ρ))

min

:

As may be inferred from Table 3.1, the maximum probability of the success of the loan

improves to 90.5% in the case of any term being 10 times as short. Conversely, it reduces to

45 millionths in the case of any term being 10 times as long, or to practically zero! So that the

Probability Model 31

Table 3.1 Subsection 3.3.2 results

ν 0.1 1 2 10 100 1000

e

max

45ppm 36.8% 60.7% 90.5% 99.0% 99.9%

j

min

∼100% 63.2% 39.3% 9.5% 1.0% 0.1%

shortfall risk of a loan may be kept ‘small’, the number of periods in the term of the loan must

be ‘small’ in relation to the quotient 1/ρ.

Seen from the operational banking point of view, this allows us to derive a policy on the

intervals of time at which loans should be examined.

It is intended to clarify this by an example: let us assume that the risk of default on a

loan should not be greater than 0.1% up to the next loan review. According to Table 3.1 this

corresponds to a value of ν = 1000. The shortfall risk might be ρ = 0.2% per annum. That

results in 1/(ν · ρ) = 0.5, i.e. this loan must be assessed every six months.

3.4 RESULTS AND CONCLUSIONS

As we have managed to show in Section 3.2, the following correlations exist between the

probability ψ

j

, the shortfall risk ρ and the survival chance χ:

ψ

j

= ε

j

=

j

¸

k=1

χ

k

=

j

¸

k=1

(1 −ρ

k

) (3.18)

and, respectively:

ψ

j

= χ

j

= (1 −ρ)

j

if χ

1

= · · · = χ

n

= χ (3.19)

In the event that the shortfall risk ρ or the survival chance χ are known for a speciﬁed period

of time, then the values for t periods may be calculated as follows; t may be assumed to have

all values between zero and inﬁnity:

ρ(t ) = 1 −(1 −ρ)

t

(3.24)

χ(t ) = χ

t

(3.28)

Under the conclusions of Section 3.3, all loans should be limited in time or provided with a

contractual clause providing for unilateral termination. This follows fromthe fact that any loan

with an inﬁnite term will, in accordance with practical considerations, default at some point in

time. The maximum term must be selected in proportion to the quotient 1/ρ. A bank’s lending

policy will ﬁnd its expression in the value ascribed to this factor. The methods of calculation

outlined in Table 3.1 and in turn in subsection 3.3.2 formthe necessary aids to decision making

in this respect.

4

Calculation of the Shortfall Risk Hedging

Rate in the Special Case of Shortfall

Risks being Constant

This chapter is concerned with calculating the shortfall risk hedging rate ρ for different types

of loan, using equations (1.2), (2.4) and (3.19), considering for the time being just the special

case ρ

1

= · · · = ρ

n

= ρ over the entirety of n periods of the loan term.

It is completely clear that the assumption of the shortfall risk ρ being constant over the

entire term of any loan is unrealistic. This assumption will be dropped in Chapter 5, and we

will show how the general case of shortfall risks being non-constant may be combined with

the results of Chapter 4. This course of action was chosen in order to make the derivation of

the results more open. That shortfall risks are not in fact constant over time will be shown in

Section 7.5 (cf. Figures 7.10 and 7.11).

Sections 4.1 and 4.6 are central to this chapter. The correlation between the shortfall risk

hedging rate r, the breakdown distribution rate probability b, the shortfall risk ρ and the risk

exposure ρ

∗

will be derived in Section 4.1. We will show in Sections 4.2 to 4.4 that all known

forms of clean credit may be described substantially with the same equations as in Section 4.1,

and that assessment of loans follows the same rules.

Based on Section 4.1, some operational conclusions are derived in Section 4.6. Section 4.7

summarises the most important results again, and these are illustrated in Section 4.8 by means

of an example.

4.1 FIXED ADVANCE WITHOUT REPAYMENTS

Loans of this kind take the form, at ﬁrst, of a cash ﬂow from the bank to the borrower, i.e. the

total of the loan is paid out. There then followseveral cash ﬂows fromthe borrower to the bank,

covering the regular interest payments and the repayment of the loan at the end of the term. In

the event that the borrower defaults, the breakdown distributions concerned take the place of

repayment. If the loan is covered, the equivalent values should count as being included in the

collateral.

The following thus applies, according to equations (1.2), (2.4) and (3.19):

Λ =

n

¸

j =1

χ

j

· i · L

(1 +i

s

)

j

+

χ

n

· L

(1 +i

s

)

n

+

n

¸

j =1

χ

j −1

· ρ · b · L · (1 +i )

(1 +i

s

)

j

(4.1)

Λ : loan market value

34 Risk-adjusted Lending Conditions

L represents the amount of loan paid out. The ﬁrst summand (in the ﬁrst set of brackets)

represents the sum of the discounted expectation values of the interest payments. The second

summand represents the expectation value of the discounted loan repayment.

The third summand (in the second set of brackets) represents the sum of the expectation

values of the breakdown distributions for each individual period. Here it has been assumed

that the breakdown distributions will be paid out at the end of the period in which the borrower

defaulted and will always be of the same size, irrespective of the period concerned. The factor

χ

j −1

represents the probability that default on the loan will not have occurred in the ﬁrst j −1

periods.

The factor ρ represents the probability that default on the loan does occur in period j .

L(1 +i ) represents the lender’s demands, where for the sake of simplicity it is assumed that

the full amount of interest for the period j will become due in the case of bankruptcy. It

is thus being assumed, for the sake of simplicity and without departing much from real-

life circumstances, that if bankruptcy occurs at all it occurs precisely at the end of the pe-

riod concerned. Under our assumptions interest was in practice paid during the ﬁrst j −1

periods.

Comparison with the study of Fooladi, Roberts and Skimmer is interesting at this juncture.

Their thesis is indeed the duration of obligations in cases where credit-worthiness is at risk

[FRSK97]; they did, however, also have to establish a correlation between loan interest,

risk-free interest and shortfall risk. Their starting equation is therefore very similar. They

do, however, deal with the general case right from the beginning (see later in Chapter 5).

When a loan is paid out, its market value must correspond at least to its nominal value:

Λ ≥ L. The bank would otherwise be already accepting a loss at that point. In what follows it

is intended that r is calculated in such a way that the market value and nominal value of the

loan are identical when it is paid out: Λ = L (cf. Section 1.6).

The result reads as follows (see Appendix 5 for derivation):

r =

ρ

∗

1 −ρ

∗

· (1 +i

s

) (4.14)

In this way the ﬁnal result for the shortfall risk hedging rate r is, following fromthe assumption

that shortfall risks are constant, independent of the number of periods in the term of the

loan!

This comes about because the revenue arising from the shortfall risk hedging rate in the

case of interest being paid in each period precisely covers the rise in the shortfall risk of the

loan repayment. Seen mathematically, it is a consequence of the properties of geometrical

series.

The correlation between r, i

s

, and ρ

∗

and b is portrayed graphically in Figures 4.1 and 4.2.

The higher the risk-free rate of interest i

s

, the higher will also be the risk premium r in

the case of the same credit shortfall risk ρ

∗

, as interest payments do indeed also have to be

‘insured’ from the point of view of risk.

If a 100% breakdown distribution were assumed in an extreme case, then the risk of default

on the loan would be zero, and with it the risk premium too.

Constant Shortfall Risk 35

Figure 4.1

b

r

r = 1%, i

s

= 3%

1%

0.5%

0%

0% 50% 100%

Figure 4.2

36 Risk-adjusted Lending Conditions

4.2 FIXED ADVANCE WITH REGULAR REPAYMENTS

This type of loan relies on ﬁxed advance disbursement without repayments, in that each

individual repayment tranche is regarded as an independent loan in the form of a ﬁxed advance

disbursement, without repayments. Equation (4.14) applies again to these individual loans. The

same shortfall risk hedging rate results for all the part loans, on account of their independence

from the number of periods in the term of the loan, and therefore for the loan as a whole too.

4.3 LOANS ON REGULAR ANNUAL REPAYMENT

In Swiss banking practice, consumer credit and leasing credit is mainly granted in the form of

loans on regular annual repayment. The borrower or lessee thus makes regular ﬁxed payments

(for example monthly, quarterly or every six months) to the lender or lessor, as the case may

be. The amount includes both interest and repayment. As the amount is ﬁxed, it contains a

high proportion of interest and a low proportion of repayment at the beginning of the term.

Towards the end of the term it is exactly the other way round, as the proportion of interest in

the total annual payment becomes smaller and smaller owing to the repayments accumulating

(cf. [BRMY96, S. 39–41].

The same considerations apply analogously to this type of credit as to ﬁxed advance with

regular repayments. The only difference between these two types of credit lies in the fact that

the sizes of the ‘imagined’ part loans vary considerably. This does not, however, alter the

validity of equation (4.14).

4.4 CURRENT ACCOUNT CREDIT

In the case of current account credit, the bank agrees a maximumcredit limit with the borrower.

In contrast to ﬁxed advance credit this gives the borrower leeway to decide each day howmuch

credit to take up, as he draws money in or pays it out. Loan interest is only calculated on the

amount of credit actually utilised each day.

The following problems for the bank in respect of the three elements that make up the loan

interest rate (see equation (1.1)) arise from this.

Financing Rate f

No problems arise here if we make the assumption that only those costs which are incurred in

proportion to the amount of credit taken up are built into the ﬁnancing rate f .

Proﬁt Contribution Rate p

Granting and supervising current account credit creates costs for the bank which are incurred

anyway, whether or not the credit is then taken up. The best case for the bank is the permanent

and full utilisation of the credit, as this allows the highest amounts of interest, and thus the

highest contributions to the covering of costs, to be billed. The worst case is, vice versa, when

the credit is granted but not utilised, as in this case there are indeed costs, but no service is

Constant Shortfall Risk 37

provided that can be billed. In accordance with Section 1.2, however, it is not our intention to

go into this problem more closely here.

Shortfall Risk Hedging Rate r

The situation here is similar to the proﬁt contribution rate. Revenue from the shortfall risk

hedging rate only arises if the credit is taken up. Vice versa, however, risk for the bank only

arises for the bank if it is taken up and only to the extent that it is utilised. So no problem arises

as long as the average utilisation of the loan prior to default on it is the same as the extent to

which the credit is taken up at the time of the default.

Banking practice does, however, paint a different picture. Experience shows that current

account credit is fully utilised at the time of default, and from time to time is even actually

exceeded. On the other hand, by virtue of the very nature of current account credit, utilisation

is by no means always necessarily full prior to default. Even borrowers who do not default

do not always use their credit limit absolutely fully and thus never pay, in relation to the limit

granted (which corresponds de facto to the bank’s effective lending risk), a full risk premium.

The circumstances outlined have the effect that equation (4.14) may not be applied across

the board in the case of current account credit. The overall necessary return from the shortfall

risk hedging rate R

n

across a bank’s entire current account credit portfolio of m individual

credit arrangements is calculated by the summation of all m individual loans at the shortfall

risk hedging rate ρ

∗

j

concerned, according to equation (4.14), multiplied by the loan granted

in each case L

g

(see Appendix 5 for derivation):

R

n

=

m

¸

j =1

ρ

∗

j

1 −ρ

∗

j

· (1 +i

s j

) · L

gj

(4.19)

R

n

= necessary return

L

g

= granted loan

But only the current account credit write-off risk hedging rate r

cacj

, agreed with the customer,

multiplied by the average amounts of credit taken up in each case L

u

, may be billed:

R

b

=

n

¸

j =1

r

cacj

· L

uj

(4.20)

R

b

= billed return

r

cac

= current account credit write-off risk hedging rate

L

u

= used loan

The simplest solution for bridging this difference consists in demanding an arrangement fee

for eachcurrent account credit, whether or not the credit is takenup. The size of this arrangement

fee corresponds here to the shortfall risk hedging rate under equation (4.14), multiplied by the

loan granted. Traditionally the Swiss lending market has, however, not permitted this way of

proceeding, which is why the solution given below has been developed.

It is intended to assume in what follows, for the sake of simplicity, that in the case of current

account borrowers the bank multiplies the risk rate under equation (4.14) for all borrowers by

38 Risk-adjusted Lending Conditions

a correction factor r

c

:

r

cacj

= r

c

·

ρ

∗

j

1 −ρ

∗

j

· (1 +i

s j

) (4.21)

r

c

= correction factor

Using this correction factor, the intention is to achieve a position in which the risk premium

that is necessary and the risk premium that has been billed are identical.

The following equation for determining r

c

may now be drawn up, with the aid of equations

(4.19), (4.20) and (4.21):

m

¸

j =1

ρ

∗

j

1 −ρ

∗

j

· (1 +i

s

) · L

gj

=

m

¸

j =1

r

c

·

ρ

∗

j

1 −ρ

∗

j

· (1 +i

s

) · L

uj

(4.22)

As the correction factor is a constant, it can be placed in front of the sum. Subsequent

division gives the following result for r

c

:

r

c

=

m

¸

j =1

ρ

∗

j

1−ρ

∗

j

· (1 +i

s

) · L

gj

m

¸

j =1

ρ

∗

j

1−ρ

∗

j

· (1 +i

s

) · L

uj

(4.23)

In relation to the past, the resulting calculation can be undertaken without further ado, though

only with considerable computation. The calculation of r

c

for future periods is, however, more

problematic, as in this case the amounts of credit utilised L

uj

have to be estimated for the

future. In times of rapid economic change this might only be possible very imprecisely.

Introducing a rating system, as explained in Chapter 2, allows us to come to grips with this

difﬁculty more successfully. The imputed ρ

∗

k

for all loans within a deﬁned rating level k are

identical and therefore constant. The ﬁgures may therefore be taken out of their brackets as

follows:

r

ck

=

ρ

∗

j

1−ρ

∗

j

· (1 +i

s

) ·

m

k

¸

j =1

L

gkj

ρ

∗

j

1−ρ

∗

j

· (1 +i

s

) ·

m

k

¸

j =1

L

ukj

(4.24)

which is reduced to:

r

ck

=

m

k

¸

j =1

L

gkj

m

k

¸

j =1

L

ukj

(4.25)

The risk premium rate r

cack

for any current account credit at the rating level k with m

k

individual loans and the credit shortfall risks therefore associated with them of ρ

∗

k

is thus

Constant Shortfall Risk 39

calculated in the following way:

r

cack

=

m

k

¸

j =1

L

gkj

m

k

¸

j =1

L

ukj

·

ρ

∗

j

1 −ρ

∗

j

· (1 +i

s

) (4.26)

The values for the future periods to be calculated must be inserted in the case of both totals

in equation (4.26). In practice it remains to be investigated whether the future values may be

approximated sufﬁciently precisely by reference to past values, or whether a special procedure

has to be developed.

High unutilised current account credit limits lead to high rates of interest on current accounts,

because of the quotient of the two totals in equation (4.26). This results in so-called ‘inventory

limits’ (i.e. longer term unutilised current account credit limits) inevitably raising the level

of current account interest rates. ‘Inventory limits’ are thus to be avoided, if possible, in the

context of being competitive as far as interest rates are concerned. For the borrower this means

that it should be budgeting its future credit needs as precisely as possible and applying to

its bank for an appropriate credit limit. The more accurately it is in a position to budget, the

less need it has to bring imponderables into its considerations. This has the consequence that

borrowers should not apply to banks for unnecessarily high credit limits just because, for

example, a competitor has a similar credit limit at its disposal, or because the borrower wishes

to make capital out of the size of its credit limit.

The problem of ‘inventory limits’ is at its most conspicuous with construction loans. By

their very nature these are only utilised fully at the end of the term, i.e. prior to consolidation.

If a construction loan deteriorates from the point of view of credit-worthiness during the

construction period —whether it be owing to the client’s ﬁnancial standing or to the quality

of the construction project in hand —it is usually pointless for the bank to call in the loan, as

during the termof such a loan there is only an unﬁnished building to showfor it. In this situation

the bank has no alternative but to complete the building at its own expense and therefore in

practice, de facto, to pay the loan off. As invoices for works completed are in each case settled

as late as possible, the extent to which any construction loan is taken up is equivalent to an

average utilisation of about one-third. Under equation (4.26) this leads to a shortfall risk charge

three times higher than that for full utilisation. We permit ourselves at this point the conjecture

that construction loans, seen in isolation, have historically been under-priced. Attractively

priced construction loans may, however, be a means of canvassing new mortgage business,

and therefore be justiﬁed on marketing grounds, but this presupposes a matching degree of

customer loyalty.

As already mentioned the problem of only partial utilisation, particularly in the case of

current account loans, may also be solved by arrangement fees on credit limits granted. This

would also be a solution in relation to the billing of proﬁt contributions that might not otherwise

be obtained. Furthermore the difﬁculty of ‘inventory limits’ could be effectively countered, as

presumably only a very fewborrowers are prepared to pay an arrangement fee for a credit limit

that may hardly ever be utilised. If the price elements p and r (see equation (1.1)) are billed in

the form of an arrangement fee, then on the other hand it is still only the price element f that

40 Risk-adjusted Lending Conditions

rises in proportion to the loan being taken up. In practice this may, for example, lead to a current

account credit priced today with interest at 5% per annum and with a fee of 0.25% per quarter

being demanded, being charged in future at 4% per annum on the amount of loan actually

taken up and at a fee of 0.5%per quarter on the amount of credit granted. Such an arrangement

fee should not, however, be forced upon the Swiss market for credit at the moment.

If cases of exceeding limits are ignored, it may be said that the following always applies:

L

uk

< L

gk

(4.27)

From this it may be concluded that the following always applies:

r

ck

> 1 (4.28)

This allows us to note, purely qualitatively, that any current account credit facility for the

same borrower is always associated with a higher loan interest rate than ﬁxed advance credit,

providing the risk-free rate of interest i

s

is identical in both cases. But even if i

s

is not identical,

then it may be emphasised that current account credit must, on the strength of the expositions

in this section, tend to be more expensive than ﬁxed advance credit.

So a lending policy could consist in only granting ﬁxed advances that are fully utilised at any

one time and operating current accounts only on the basis of credit balances. The advantage of

being able to give immediate notice on current account facilities in cases indicating that those

facilities should be withdrawn is in many cases illusory anyway. Moreover the qualifying time

until maturity in the case of ﬁxed advances, of three or six months, would not be excessively

long. If winding up such advances does occur, one may always fall back on current account

lending. This is no arithmetical problem here, as experience shows that current account credit

facilities are fully utilised in such situations and therefore are tantamount to de facto ﬁxed

advances.

One further problem in current account lending is the tiresome subject of current account

facilities being exceeded. The most effective method of meeting this difﬁculty is the charging

of fees or of a special rate of interest on the excess, which would cover the additional risks

even in the most unfavourable instance. As the allowing of credit to be exceeded means a great

and in no way automatic concession by the bank, matching prices may also be applied here.

The exceeding of credit limits is, however, usually also an indication of deteriorating credit-

worthiness on the part of the borrower, and calls therefore for reassessment of the lending. Here

it must also be established, in particular, whether or not the high loan interest rates associated

with the excess borrowing can be borne by the borrower: over-mechanical applications of

higher rates of interest may cause additional credit-worthiness problems and thereby become

self-fulﬁlling prophecies. If credit limits are exceeded over a longer period of time, a decision

must be taken as to whether the normal credit limit could be raised, or whether it is a case for

withdrawal of credit.

4.5 LOAN ASSESSMENT

This chapter has so far shown how the shortfall risk hedging rate r should be calculated when

paying out a loan. Any borrower’s credit-worthiness may, however, change during the term of

Constant Shortfall Risk 41

a loan. By applying equation (4.1) the market value of a loan L

l

for the last l periods of the

term can be detailed as follows:

Λ

l

=

l

¸

j =1

χ

j

l

· i · L

(1 +i

sl

)

j

+

L

(1 +i

sl

)

l

+

l

¸

j =1

χ

j −1

l

· ρ

l

· b · L · (1 +i )

(1 +i

sl

)

j

(4.29)

The value of i

sl

at the time of the assessment does not necessarily have to coincide with the

value of i

s

at the time of paying out the loan, which is expressed by the index I .

The equation factor λ for correcting the assessment of market value in relation to nominal

value is derived in Appendix 5.

λ =

(1 +i

sl

)

l

−χ

l

l

· ((1 +i ) · (ρ

l

· b +χ

l

) −i

sl

−1)

(1 +i

sl

)

l

· (1 +i

sl

−χ

l

)

(4.38)

The left-hand bracket in the numerator and the denominator in equation (4.38) are always

positive because of the deﬁnitions of the values occurring. An appreciation proﬁt thus exists

if the right-hand bracket in the numerator of equation (4.38) is positive. If it is negative,

circumstances have arisen requiring a provision to be made:

(1 +i ) · (ρ

l

· b +χ

l

) > 1 +i

sl

⇒appreciation proﬁt

(1 +i ) · (ρ

l

· b +χ

l

) < 1 +i

sl

⇒requiring a provision

(4.39)

No assessment correction is necessary if this bracket is equivalent to zero, i.e.

(1 +i ) · (ρ

l

· b +χ

l

) = 1 +i

sl

(4.40)

(1 +i ) =

1 +i

sl

ρ

l

· b +1 −ρ

l

(4.41)

i =

1 +i

sl

1 −ρ

l

· (1 −b)

−1 (4.42)

i =

1 +i

sl

−1 +ρ

l

· (1 −b)

1 −ρ

l

· (1 −b)

(4.43)

i =

i

sl

+ρ

l

· (1 −b)

1 −ρ

l

· (1 −b)

(4.44)

i =

i

sl

+ρ

∗

l

1 −ρ

∗

l

= i

l

(4.45)

Equation (4.45) is identical to equation (4.49) (see Section 4.6). There is thus no need for

assessment correction if the risk-adjusted rate of interest i

l

l periods prior to the end of the term

is identical to the risk-adjusted rate of interest when the loan is paid out!

4.6 CONCLUSIONS

The following conclusions can be drawn by derivation from equation (4.14).

42 Risk-adjusted Lending Conditions

4.6.1 Minimum Loan Interest Rate

Equations (1.3) and (4.14) inserted into equation (1.1) will yield:

i = i

s

+

ρ

∗

1 −ρ

∗

· (1 +i

s

) (4.46)

Equation (4.46) can be simpliﬁed as follows:

i =

i

s

· (1 −ρ

∗

) +ρ

∗

· (1 +i

s

)

1 −ρ

∗

(4.47)

i =

i

s

−i

s

· ρ

∗

+ρ

∗

+i

s

· ρ

∗

1 −ρ

∗

(4.48)

i =

i

s

+ρ

∗

1 −ρ

∗

(4.49)

The loaninterest rate under equation(4.49) cannowbe interpretedinthe case of the minimum

rate of interest given i

smi n

or risk-free loans and the credit shortfall risk given ρ

∗

as minimum

rate of interest i

min

, and it is this loan interest rate that must be billed to the customer in order

to achieve the necessary minimum proﬁt contribution:

i

min

=

i

smin

+ρ

∗

1 −ρ

∗

=

f + p

min

+ρ

∗

1 −ρ

∗

(4.50)

See Figure 4.3 for illustration of the correlations.

i

20%

10%

0%

0% 10% 20%

5%

3%

1%

Parameter: i

s

Figure 4.3

Constant Shortfall Risk 43

4.6.2 Effective Proﬁt Contribution Rate

The bank will in practice agree with the customer an effective loan interest rate i

eff

, which is not

necessarily identical to the minimum loan interest rate i

min

according to equation (4.50). This

leads to an effective proﬁt contribution rate p

eff

, which is thus not identical to the minimum

proﬁt contribution rate p

min

and can be derived as follows from equation (4.50):

i

eff

=

f + p

eff

+ρ

∗

1 −ρ

∗

(4.51)

i

eff

· (1 −ρ

∗

) = f + p

eff

+ρ

∗

(4.52)

p

eff

= i

eff

· (1 −ρ

∗

) − f −ρ∗ (4.53)

As is discernible from equation (4.53), p

eff

may also assume negative values, and indeed if

that happens, when i

eff

is so small, that means:

p

eff

< 0 if i

eff

<

f +ρ

∗

1 −ρ

∗

(4.54)

Furthermore sensitivity analysis of p

eff

versus i

eff

∂p

eff

∂i

eff

= 1 −ρ

∗

(4.55)

shows that, when the value of i

eff

changes the value of p

eff

does not change to the full extent of

the value of i

eff

, but only by the reduced factor of 1 −ρ

∗

. That is because, in the case of a rise

in the value of i

eff

, this rise must also be covered again via a rise in the shortfall risk hedging

rate, and vice versa. This will become clear in the following subsection.

4.6.3 Effective Shortfall Risk Hedging Rate

By analogy with the effective proﬁt contribution rate, the effective shortfall risk hedging rate

can be derived by using equation (4.14):

r

eff

=

ρ

∗

1 −ρ

∗

· (1 + f + p

eff

) (4.56)

Bringing equation (4.53) into use results in:

r

eff

=

ρ

∗

1 −ρ

∗

· (1 + f +i

eff

· (1 +ρ

∗

) − f −ρ

∗

) (4.57)

r

eff

=

ρ

∗

1 −ρ

∗

· ((1 −ρ

∗

) +i

eff

· (1 −ρ

∗

)) (4.58)

r

eff

= ρ

∗

· (1 +i

eff

) (4.59)

Because ρ

∗

and i

eff

are always positive ﬁgures, it is always the case that:

r

eff

> 0 (4.60)

44 Risk-adjusted Lending Conditions

The sensitivity analysis of r

eff

in relation to i

eff

,

∂r

eff

∂i

eff

= ρ

∗

(4.61)

shows that the value of r

eff

, in the case of any change in the value of i

eff

likewise does not

change to the same extent as i

eff

, but this time changes by the reduced factor ρ

∗

.

The sum of both partial differentials equations (4.55) and (4.61)

∂p

eff

∂i

eff

+

∂r

eff

∂i

eff

= 1 −ρ

∗

+ρ

∗

= 1 (4.62)

shows, as was to be expected, that any change in the value of i

eff

is distributed completely over

the values of p

eff

and r

eff

, and indeed in the ratio

p

eff

r

eff

=

1 −ρ

∗

ρ

∗

(4.63)

As ρ

∗

is substantially smaller than (1 −ρ

∗

) in the case of borrowers of good ﬁnancial

standing, the essentially larger part of any change in the effective rate of interest on the

effective proﬁt contribution ratio does not apply.

4.6.4 Maximum Shortfall Risk Covered

In some Swiss cantons the maximum loan interest rate i

max

for consumer credit is laid down

by law. The question thus arises —up to what maximum credit shortfall risk ρ

∗

may consumer

credit be granted in the case of a given minimum risk-free standard rate of interest I

smin

? The

following applies, using equation (4.49):

i

max

=

i

smin

+ρ

∗

max

1 −ρ

∗

max

(4.64)

Solving, using ρ

∗

max

, results in:

ρ

∗

max

=

i

max

−i

smin

1 +i

max

=

i

max

− f − p

min

1 +i

max

(4.65)

Equation (4.67) makes it clear that higher values for i

max

permit higher values for ρ

∗

max

too.

Moreover it is that case that:

ρ

max

= 0 if i

max

= f + p

min

(4.66)

For the legislature laying down the value of i

max

, the necessity arises from this that:

i

max

f + p

min

(4.67)

The market for consumer credit would otherwise be reduced to vanishing point owing to

high credit-worthiness requirements.

Constant Shortfall Risk 45

4.7 RESULTS AND CONCLUSIONS

The shortfall risk hedging rate is normally calculated from the following equation according

to Section 4.2:

r =

ρ

∗

1 −ρ

∗

· (1 +i

r f

) (4.14)

In the case of current account lending a correction factor r

c

must be introduced, depending

on the average utilisation of the loan:

r

cacj

= r

c

·

ρ

∗

j

1 −ρ

∗

j

· (1 +r

scac

) (4.21)

We refer to the expositions in Section 4.5 for calculation of the correction factor.

The minimum loan interest rate is calculated from the equation:

i

min

=

i

smin

+ρ

∗

1 −ρ

∗

(4.50)

This is one of the most signiﬁcant equations. It is particularly important here to note the

numerator. The greater the shortfall risk hedging rate, the more the numerator in equation

(4.49) makes itself noticeable: the minimum loan interest rate is not simply the sum of the

minimum risk-free standard rate of interest and of risk!

After the minimum rate of loan interest i

min

has been laid down for a loan transaction, it will

in practice be rounded up to the next normal round ﬁgure (usually in the form of one-quarter,

one-eighth or one-sixteenth of 1%). This results in the effective loan interest rate i

eff

. Thus the

effective proﬁt contribution rate and the effective shortfall risk hedging rate are calculated as

follows:

p

eff

= i

eff

· (1 −ρ

∗

) − f −ρ

∗

(4.53)

r

eff

= ρ

∗

· (1 +i

eff

) (4.59)

There is an important observation here, that the effective proﬁt contribution rate becomes

negative under the condition:

i

eff

<

f +ρ

∗

1 −ρ

∗

(4.54)

The effective shortfall risk hedging rate may not, however, ever be negative!

4.8 EXAMPLE

It is intended to grant a borrower a ﬁxed advance for one year. The starting position looks like

this:

46 Risk-adjusted Lending Conditions

Financing cost rate: f = 3.000%

Minimum proﬁt contribution rate: p

min

= 1.000%

Credit shortfall risk: ρ

∗

= 0.757%

(Rating BB according to Table 2.1)

The minimum risk-free standard rate of interest is:

i

smin

= 3%+1% = 4%

The shortfall risk hedging rate, according to equation (4.14) is:

r =

0.00757

1 −0.00757

· (1 +0.04) = 0.7933%

From this is calculated the minimum loan interest rate according to equation (1.1) by

i

min

= 4%+0.7933% = 4.7933%

One obtains the same result, moving directly, via the shortfall risk hedging rate r under

equation (4.49):

i

min

=

0.04 +0.00757

1 −0.00757

= 4.7933%

The ﬁgure of 4.7933% is not customary in Swiss banking practice. It is therefore rounded

up to the nearest eighth:

i

eff

= 4

7

8

% = 4.875%

The effective proﬁt contribution rate and the effective shortfall risk hedging rate are calcu-

lated, with the aid of equations (4.53) and (4.59), as follows:

p

eff

= 0.04875 · (1 −0.00757) −0.03 −0.00757 = 1.0811%

r

eff

= 0.00757 · (1 +0.04875) = 0.7939%

The proof is:

i

eff

= 3%+1.0811%+0.7939% = 4.875% = 4

7

8

%

5

Calculation of the Shortfall Risk Hedging

Rate in the General Case of Variable

Shortfall Risk

We now give up, in this chapter, the condition ρ

1

= · · · = ρ

n

= ρ. First the exact solution

for the case of a ﬁxed interest loan without repayments will be derived in Section 5.1. In

Section 5.2 we will draw up, using an approximate solution as an example, a comparison with

the special case in Chapter 4, and we will examine the accuracy of the approximate solution in

Section 5.3.

Other important clean credits will be dealt with in Sections 5.4, 5.5 and 5.6. The important

results and conclusions from Chapter 5 will be summarised in Section 5.7.

The aim of this chapter is to elaborate general principles, which will then be developed

further in Chapter 7.

5.1 FIXED INTEREST LOAN WITHOUT REPAYMENTS

First it is intended to examine again, as in Chapter 4, the ﬁxed advance without repayments.

Equation (3.18) must, however, now be brought into play for the value ψ

j

in equation (1.2).

By analogy with equation (4.1) this results in (for cases where the loan is covered, the values

are on the other hand valid when the collateral is included):

Λ =

¸

¸

¸

¸

n

¸

j =1

i · L ·

j

¸

k=1

(1 −ρ

k

)

(1 +i

s

)

j

+

¸

¸

¸

L ·

n

¸

k=1

(1 −ρ

k

)

(1 +i

s

)

n

+

¸

¸

¸

¸

n

¸

j =1

b · L · (1 +i ) · ρ

j

·

j −1

¸

k=1

(1 −ρ

k

)

(1 +i

s

)

j

(5.1)

The left-handsummandagaincorresponds tothe discountedexpectationvalues of the interest

payments, the centre summand to the expectation value of the loan repayment and the right-

hand summand to the sumof the discounted expectation values of the breakdown distributions.

It will again be assumed, as in Chapter 4, that there is one uniform risk-free rate of interest i

s

for all terms. Λ = L will again be set (see Section 4.1) for the period in which the loan is paid

out.

48 Risk-adjusted Lending Conditions

The following substitutions will be made in order to be make handling equation (5.1) more

manageable:

x =

n

¸

i =1

j

¸

k=1

(1 −ρ

k

)

(1 +i

s

)

j

(5.2)

y =

n

¸

k=1

(1 −ρ

k

)

(1 +i

s

)

n

(5.3)

z =

n

¸

j =1

ρ

j

·

j −1

¸

k=1

(1 −ρ

u

)

(1 +i

s

)

j

(5.4)

With x, y and z brought into play, and abbreviated with L, there results:

1 = i · x + y +b · (1 +i ) · z (5.5)

and solved using i :

i =

1 −(y +b · z)

x +b · z

(5.6)

Logically, no further simpliﬁcation of the representation can be achieved by any reverse

substitution. Equations (5.2), (5.3), and (5.4) should therefore be seen as computational indi-

cations, the results of which may be brought into play in equation (5.6). These equations can

be used with appropriate PC worksheets without any difﬁculty.

The mean shortfall risk ρ

m

, which implicitly underlies the interest rate calculated in this

way, is calculated as follows using equation (4.67) as an example:

ρ

m

=

i −i

s

1 +i

(5.7)

5.2 APPROXIMATE SOLUTION FOR FIXED INTEREST LOAN

WITHOUT REPAYMENTS

The computing needs for Section 5.1 are in practice indeed applicable, but very unwieldy. An

approximate solution is therefore derived in this section.

The shortfall risk ρ

k

of the k

th

period can be replaced by the sum of the average shortfall

risk ρ

a

of all n periods and the k

th

deviation ρ

k

by this means:

ρ

k

= ρ

a

+ρ

k

(5.8)

with ρ

a

=

n

¸

k=1

ρ

k

n

Variable Shortfall Risk 49

Equations (5.2) for x, (5.3) for y and (5.4) for z now read:

x =

n

¸

j =1

j

¸

k=1

(1 −[ρ

a

+ρ

k

])

(1 +i

s

)

j

(5.9)

y =

n

¸

k=1

(1 −[ρ

a

+ρ

k

])

(1 +i

s

)

n

(5.10)

z =

n

¸

j =1

(ρ

a

+ρ

j

) ·

j −1

¸

k=1

(1 −[ρ

a

+ρ

k

])

(1 +i

s

)

j

(5.11)

The approximate solution is derived in Appendix 6 using conventional approximation equa-

tion. The result is:

i =

i

s

+ρ

∗

a

1 −ρ

∗

a

(5.29)

Comparison with equation (4.49) shows that it is identical to equation (5.29). In the case of

the variable ρ

j

equation (4.49) may thus be used, likewise approximately, in that the constant

ρ

∗

is replaced by the average ρ

∗

a

. On the basis of the type and method of derivation used, this

result could indeed have been expected.

5.3 RELIABILITY OF THE APPROXIMATE SOLUTION

At this point the reliability of the approximate solution will be illustrated by means of six

numerical examples. Let’s put i

s

= 5%, ρ

∗

a

= 1% and b = 0 in all the examples. The approxi-

mate solution is thus always i = 6.0606%. The values for the shortfall risk ρ concerned in the

individual years may be taken fromTable 5.1. The value of i , calculated precisely, is detailed in

the third column from the right-hand side and may be compared with the approximate solution

i = 6.0606%.

The examples were selected as follows:

r

a and d represent deteriorating credit-worthiness

r

b and e represent improving credit-worthiness

Table 5.1 Reliability of the approximate solution

Values for ρ

k

year Results

1 2 3 4 5 i

exact

(%) abs. error rel. error

a 0.8% 0.9% 1.0% 1.1% 1.2% 6.0481% 0.0125% 0.2062%

b 1.2% 1.1% 1.0% 0.9% 0.8% 6.0733% −0.0127% −0.2092%

c 1.0% 1.1% 1.0% 0.9% 1.0% 6.0631% −0.0025% −0.0418%

d 0.6% 0.8% 1.0% 1.2% 1.4% 6.0359% 0.0247% 0.4093%

e 1.4% 1.2% 1.0% 0.8% 0.6% 6.0863% −0.0257% −0.4215%

f 1.0% 0.8% 1.0% 1.2% 1.0% 6.0557% 0.0049% 0.0816%

50 Risk-adjusted Lending Conditions

r

c and f represent oscillating credit-worthiness

r

in c there is deterioration at ﬁrst and in f there is improvement at ﬁrst

r

in a, b and c the changes in each case are half the size of those in d, e and f.

The following can be seen from the results:

r

The approximate value for i is too high in the case of deteriorating credit-worthiness, and

too low in the case of improving credit-worthiness.

r

In the case of oscillating credit-worthiness it depends whether it has improved at ﬁrst

(approximate solution too high) or deteriorated ﬁrst (approximate solution too low).

r

The deviation of the approximate solution is about double in the case of changes that are

double the size.

r

Although great ﬂuctuations in credit-worthiness were inserted into the examples, the absolute

error amounts to only a few basis points or fractions of a basis point.

As has already been emphasised in Section 5.1, it is possible to use the exact equations with

appropriate PC worksheets. In most cases, however, usable results can also be delivered by

approximate solution.

5.4 FIXED ADVANCE WITH COMPLETE REPAYMENT

Equation (1.2) appears in this case as follows:

Λ =

n

¸

j =1

×

¸

i · L ·

¸

1 −

( j −1)

n

¸

+

L

n

¸

·

j

¸

k=1

(1 −ρ

k

) +b · ρ

j

·

¸

(1 +i ) · L ·

¸

1 −

( j −1)

n

¸¸

·

j −1

¸

k=1

(1 −ρ

k

)

(1 +i

s

)

j

(5.30)

The ﬁrst summand in the numerator corresponds to the expectation values of the interest

payments and repayments. The ﬁrst summand in the left-hand curved brackets corresponds to

the interest payments, with the square brackets detailing the rest of the loan subject to interest

in each case. The second summand in the left-hand curved brackets corresponds to the tranche

of repayment. The second summand in the numerator corresponds to the expectation values of

the breakdown distributions. The square brackets in the right-hand curved brackets represents

the remaining debt. The other terms are analogous to those in equation (5.1).

The result for i is derived in Appendix 6 on the assumption that Λ = L at the time of paying

out. The result is:

i =

n

n

¸

j =1

¸

1 +

b · ρ

j

1−ρ

j

·

1 +

1

i

· (n − j +1) +

1

i

¸

·

j

¸

k=1

(1 −ρ

k

)

(1 +i

s

)

j

(5.38)

Variable Shortfall Risk 51

Table 5.2 Interest rate comparison for loans with and

without repayments

Example i

without amo

i

with amo

i

with

less i

without

a 6.0481% 5.9798% −0.0683%

b 6.0733% 6.1418% 0.0685%

c 6.0631% 6.0768% 0.0137%

d 6.0359% 5.8994% −0.1365%

e 6.0863% 6.2234% 0.1371%

f 6.0557% 6.0284% −0.0273%

The rate of interest i in equation (5.38) occurs both to the left and to the right of the equals

sign. In this format the equation is therefore soluble iteratively with the aid of an appropriate

PC worksheet.

In order to illustrate equation (5.38), the interest rates for ﬁxed advances with repayments in

Table 5.2 are calculated with the risks over term examples of Table 5.1 and compared with the

rate of interest of ﬁxedadvances without repayments. The standardrate of interest i

s

is again5%.

It is nowinteresting to realise that lower interest rates result only in examples a, d and f in the

case of repayments. These are the cases of deteriorating credit-worthiness. Here repayments

clearly have the effect of reducing risk. In the cases of improving credit-worthiness it is the

other way round. In these cases, because lower amounts of loan are bearing interest on account

of repayments in periods of good credit-worthiness, only correspondingly lower revenues may

be billed. The result of this is that a higher margin is needed at the beginning of the term of

the loan, which leads to higher rates of interest in relation to loans without repayments.

5.5 FIXED ADVANCE WITH PARTIAL REPAYMENTS

Equation (5.23) has to be supplemented as follows for this case:

L =

n

¸

j =1

×

¸

i · L ·

¸

1 −

α · ( j −1)

n

¸

+

α · L

n

¸

·

j

¸

k=1

(1 −ρ

k

) +b · ρ

j

·

¸

(1 +i ) · L ·

¸

1 −

α · ( j −1)

n

¸¸

·

j −1

¸

k=1

(1 −ρ

k

)

(1 +i

s

)

j

+

(1 −α) · L ·

n

¸

k=1

(1 −ρ

k

)

(1 +i

s

)

n

(5.39)

Here α represents the degree of repayment with 0 < α ≤ 1. The solution path matches that

in the preceding section and leads once more to an iterative solution. We will not go into the

derivation at this point. The solution lies in each case between that for loans without repayments

and that for loans with complete repayments over the term.

52 Risk-adjusted Lending Conditions

5.6 CURRENT ACCOUNT LOANS

As interest rates on current account loans may be adjusted in the short term to new circum-

stances, there is no problem regarding changes in the borrower’s credit-worthiness. As soon

as the bank establishes such situations, it simply recalculates and adjusts the loan interest rate.

5.7 RESULTS AND CONCLUSIONS

As Tables 5.1 and 5.2 show, real-life loans may normally be handled by using the average short-

fall risk ρ

a

according to equation (5.8), in conjunction with the equations given in Chapter 4.

If it is suspected that this approximation is insufﬁciently accurate, this chapter supplies the

details needed to check the approximation or to calculate interest rates more precisely.

The results to date are only applicable as an autonomous model if the bank has the necessary

statistical data available. It also needs an operational accounting system, such as is normally

found in insurance companies, in order to be able to calculate the parameters used in the model

in relation to the past. Forecasts for the future can be drawn up supported by these and the

corresponding values inserted into the model. Chapter 6 outlines how to go about this.

Above all, however, the calculations we have made so far serve as a basis on which to develop

further the approach of Black and Scholes [BLSC73, S. 673 and thereafter] in Chapter 7.

Equations (4.49) and (5.29) form the necessary prerequisites for this.

6

Shortfall Risk on Uncovered Loans on the

Basis of Statistics

This chapter will open up opportunities for determining the shortfall risks of borrowers with

the aid of statistics. Experience in Swiss bank lending shows that loans to businesses and

to private individuals more or less balance out, both in terms of numbers and in terms of

amounts. Separate and detailed consideration of both these groups of customers is therefore

justiﬁed.

6.1 PRIVATE CLIENTS

As will be shown in Chapter 8, the shortfall risk of covered loans —as these predominantly

are in the case of private client business (mortgages, loans against other collateral) —is made

up of the shortfall risk for uncovered loans to the same borrowers and of the shortfall risk of

the cover. The shortfall risk for loans to private clients on an uncovered basis will therefore be

examined below. (Domowitz and Sartain have as it happens established in a recent study for

the USA [DOSA97], that invoices for medical services and credit card balances represent the

most frequent cause of personal bankruptcy. A development that may possibly be in store for

Switzerland too?)

Private individuals normally have four possible sources of income with which to ﬁnance

their debt servicing:

r

income from salaried employment

r

income from self employment

r

pensions (old age and dependant’s insurance, pension funds, life insurance, etc.)

r

investment income and assets

Earned income and pensions are normally mutually exclusive and may thus be considered

separately. Investment income comes, when there is any, on top of them. Any loan to a private

individual is subject to default if the person concerned no longer has sufﬁcient income available

to service the debt. The various sources of income and the effects they have on the shortfall

risk will be examined separately below.

6.1.1 Unearned Income and Income from Self-employment

The shortfall risk of the private individual in these cases corresponds to the shortfall risk of

the pension fund or of his/her own company. It will therefore be referred to in Section 6.2

instead.

54 Risk-adjusted Lending Conditions

6.1.2 Income from Salaried Employment

It is assumed in this subsection that the bank will grant a private individual an uncovered loan

amounting to L

s

on the strength of his/her salary. The size of the loan L

s

in relation to salary

is determined on the basis of the bank’s credit policy. A defensive attitude by the bank may,

in an extreme case, take the form of uncovered loans not being granted to private individuals

in principle. That would, however, also mean not granting any normal consumer credit either!

Establishment of loan amount L

s

therefore takes place in line with appropriate practice, taking

account of salary in accordance with the normal rules of consumer credit business. The size

of L

s

is thus aligned with the acceptability of the debt servicing potential.

The bank’s risk here is that the borrower becomes unemployed and remains so in the

long term. Short-term unemployment can usually be bridged with the aid of unemployment

insurance. The probability of the borrower becoming unemployed in the long term within one

year can be investigated on the basis of unemployment ﬁgures relating to his/her occupation,

age and location. From this the bank must then determine for itself whether this probability is

more or less likely to befall the borrower concerned over the next year. Economic forecasts for

the occupational or professional group concerned and for the region, as well as the personal

qualiﬁcations of the borrower, play a decisive role in this. This is obviously not easy, but the

bank nonetheless has to form a view for itself.

In addition to the risk of unemployment, the risks of invalidity and death have to be taken

into account. Appropriate policies may, however, be taken out with the insurance industry to

cover these risks, and these may be assigned in favour of the loan. We will not therefore go

further into these risks at this point.

6.1.3 Investment Income and Assets

This subsection will distinguish on the one hand between securities and credit balance assets,

and property assets on the other hand.

Revenues from securities and credit balance assets are often very volatile. The probability

of a person defaulting on an uncovered loan that is backed up solely by income from assets

must therefore be considered very high. If a private borrower only has income from securities

and credit balances available, then the shortfall risk should be calculated only on the basis of

covering the loan. Loans that a bank may make against other collateral should be evaluated

according to appropriate statistics.

The case of income arising from returns on real property is quite different. Long-term

and sustained revenue can be perfectly well achieved from real property, and can be very

precisely estimated on the basis of experience in the property sector. The risk for the bank

consists in the possibility of the management of the property in question being improperly

conducted, and of its long-term revenue thus being jeopardised. Such cases have a feature

in common and comparable with that of companies. In the case of returns from property

in private ownership, a ‘management agreement’ can be drawn up, as indeed has already

been done on occasion. This then becomes a situation for review, as will be described in

Section 6.2.

Shortfall Risk on Uncovered Loans Using Statistics 55

6.2 COMPANIES

Historically orientated deﬁnitions of the probability of a company defaulting on a loan can be

made on the basis of bankruptcy statistics. In order to arrive at the most precise ﬁgures possible

for individual companies it is necessary to classify the entirety of all companies as neatly as

possible. Here it makes sense to weigh up against each other the need for the neatest possible

classiﬁcation and the need for results that are statistically meaningful. Particular criteria for

classiﬁcation are:

r

business sectors

r

geographical/economic areas

r

sizes of enterprises by number(s) of employees

r

how long companies have been established

Such details may be obtained —for foreign companies too —and are included in every

census of business operations in Switzerland. The most intractable problem is certainly classi-

ﬁcation by sectors, as many companies clearly cannot simply be ﬁtted into just one particular

sector.

How far back the necessary period for consideration should go depends on the size of the

individual segments, and on the need to obtain statistically meaningful results. On the other

hand the observation period should not be too long, in order to have facts that are as up to date

as possible, and which provide clues that will be applicable for the following period.

Analysing bankruptcy statistics always involves reappraisal of the past but, on the other

hand, the future is far from simply being an extrapolation of it.

Banks cannot therefore avoid forming an opinion on the risks of defaults arising in indi-

vidual segments over the coming periods concerned, based on the one hand on historical (but

reappraised) facts and on the other hand on economic forecasts.

The facts behind the bankruptcy statistics should be as up to date as possible for the purposes

described above. The degree of relevance required is not necessarily provided by facts that are

made available from ofﬁcial sources. On the strength of their high market shares the major

banks, in particular, and the cantonal banks within their cantonal areas, should, however, have

sufﬁciently accurate facts at their disposal, even if they only use them for making evaluations

within their own established clientele.

We will not go any further into such statistical methods here. We refer you at this point to

appropriate specialist reading (such as [BOHL92]).

Part III

Option-Theory Loan Risk Model Option-Theory Loan Risk Model

Shortfall risk on uncovered loans to companies on the basis of an

option-theory approach

Loans covered against shortfall risk

Calculation of the combination of loans with the lowest interest costs

7

Shortfall Risk on Uncovered Loans to

Companies on the Basis of an

Option-Theory Approach

There are some hurdles to be cleared, as has been shown in Section 6.2, when it comes to the

statistical determination of the shortfall risk of uncovered loans to companies. The relevance

and necessary scope of the facts is uncertain. The individual assessment of any one company

is only possible by cross-comparison, in so far as the facts needed may exist. The largely

inadmissible extrapolation fromthe past into the future can only be avoided by using economic

forecasts, for which the appropriate facts and contexts have to be known.

This cannot be satisfactory. We therefore introduce a fundamentally different way forward

in this chapter. A company will be individually assessed, using an option-theory approach, on

the basis of operational information. As a second step it will be demonstrated, on the strength

of analogous conclusions, how loans to private individuals may be assessed too.

7.1 DIFFERENCE IN APPROACH BETWEEN BLACK/SCHOLES

AND KMV, TOGETHER WITH FURTHER ELABORATION

It was Black and Scholes who ﬁrst described the equity of a company partly ﬁnanced by

outside capital as a call-option and the debts as risk-free credit, combined with a put-option

on the total value of the company [BLSC73, S. 637 and thereafter]. In this their reﬂections

were based on the assumption that the outside capital was made available in the form of a zero

bond. Cox and Rubinstein [CORU85, S. 375 and thereafter] described this approached later,

with further reﬁnements. More recently this approach has been taken up again by Grenadier

[GREN96].

As already mentioned in Section 1.2, the KMV Corporation in San Francisco, California,

USA, is adopting a similar approach (see [VAS184] and [KAEL98]. Here an inference was

made from the volatility of the stockmarket prices of listed companies to the volatility of their

values. Their model does not, however, use the Black/Scholes formula, but pursues, starting

likewise froma stochastic process, a similar solution approach. In contrast to this, the volatility

of the capitalised free cash ﬂows will be used, in the context of this study, to assess the volatility

of values of the company, and the Black/Scholes formula will be used as a model. Loans to

non-listed companies may thereby be assessed also.

The Black/Scholes approach is developed further here, with debts no longer being examined

in the form of a zero bond. It will rather be assumed here that the repayment of the loan paid

out L plus the agreed interest at the rate of i is owed at the end of any period of time. This is

completely permissible, as the nominal value of a zero bond may also be seen as its market

60 Risk-adjusted Lending Conditions

value with interest on it accrued. Every investor is taking, as the basis for an investment in

zero bonds, a yield to be achieved that determined the relationship between market value and

nominal value. This consideration leads logically to the conclusion that the nominal value of

the zero bond may be replaced by the magnitude Λ · (1 +i ), in which Λ corresponds to the

current market value and i . corresponds to the return on the market value set by the investor

over the remainder of the term. In applying these considerations to loans, Λ = L will be put in

again, as we did in Sections 4.1 and 5.1. The nominal value X of the zero bonds according to the

notation by Black/Scholes may thus be seen as shorthand for X = L · (1 +i ). The result has

to be the same in the end. A generalised solution is, however, obtained by applying equations

(4.49) and (5.29) respectively.

As will be demonstrated in this chapter, this approach leads to a solution for the risk-adjusted

loan interest rate which, in contrast to that of Black/Scholes, is independent of the risk-free

rate of interest i

s

. It becomes possible, therefore —unlike the approach of Black/Scholes —to

assess debt structures that may be as complicated as you like, in a simple way. In Chapter 8

we will see that it is also possible to assess companies that have taken up both covered and

uncovered loans at the same time.

The fundamental of the methods being developed here will be explained ﬁrst, in the next

section. The derivation and discussion of the risk-adjusted loan interest rate at the time of

establishing the loan conditions will followin Sections 7.3 to 7.6. The intermediate assessment

of outstandingcredit will be discussedinSections 7.7and7.8, andthe inﬂuence of the privileged

wages and salaries payable in the event of bankruptcy, under Switzerland’s laws concerning the

pursuit of debt and bankruptcy, will be examined in Section 7.9. Section 7.10 will specify the

limits on the application of the method presented here, and the most important results will be

portrayed in Section 7.11. The ﬁnal section, 7.12, application of the method will be illustrated

with the aid of examples.

7.2 DERIVATION OF BASIC FORMULAE

First of all debts are assumed to be in the form of one single bank loan, in which at the end

of a period the nominal value L plus the agreed interest at the rate of i will become due for

payment according to the period concerned. The length of the period of time will ﬁrst of all be

consciously left open. This is permissible, if the rate of interest i corresponds to the length of

the period. On the basis of these assumptions the following possible values result at the end of

the period for the equity and for the debt, following on [CORU85, S. 377]:

E = Max(V − L · (1 +i ); 0) (7.1)

D = Min(V; L · (1 +i )) (7.2)

V = E + D (7.3)

E = equity

D = debts

V = value of the company

At the end of the period the market value of the company’s equity corresponds either to

the company’s market value V less the bank’s demand L · (1 +i ) —or it is, in the case of the

Shortfall Risk on Uncovered Loans: Option-Theory Approach 61

E = 0

E = V − L(1 + i )

V L(1 + i )

E

Figure 7.1

D = L(1 + i )

D = V

L(1 + i )

L(1 + i )

D

V

Figure 7.2

bank’s demand being higher than the company’s market value, equal to zero. This is expressed

in equation (7.1).

On the other hand, at the end of the term of the loan the bank receives either its demand

L · (1 +i ) paid back, or the owners of the company let it go bankrupt, as it is not worthwhile

for them to pay back a loan demand that is higher than the company’s market value. In this

situation it is more advantageous for the owners of the company to invest the money in a new

enterprise.

Debts and the market value of equity together result in the company’s value according to

equation (7.3). Figures 7.1 and 7.2 show the situation at the end of the loan’s term.

From Figure 7.1 it becomes evident that the market value of equity is no different from

the equivalent of a European call option on the company’s market value. On the other hand,

according to Figures 7.2, 7.3 and 7.4, the debts may also be portrayed as a risk-free investment

with the value L · (1 +i ) plus a European put option written on the company’s market value.

The unexpired term of these options corresponds in each case to the time until the loan falls

due [CORU85, S. 380].

On the basis of the ﬁgures it becomes furthermore clear that L · (1 +i ) corresponds to the

striking price and V to the basic value of the option. The call put parity leads to, with P

corresponding with the value of the put:

E = P + V − L · (1 +i ) · (1 +i

s

)

−1

(7.4)

62 Risk-adjusted Lending Conditions

V

L(1 + i )

L(1 + i )

−L(1 + i )

V − L(1 + i )

Figure 7.3

D = L(1 + i )

D = V

L(1 + i )

D

V

Figure 7.4

The value for the risk-free standard rate of interest i

s

must of course be put in, likewise

matching the period of time. It works out for the company’s value V as follows:

V = L ·

1 +i

1 +i

s

+ E − P (7.5)

The value of the debts works out, using equation (7.3), as follows:

D = V − E = L ·

1 +i

1 +i

s

− P (7.6)

The expression L · (1 +i ) · (1 +i

s

)

−1

is no different from the fully discounted nominal

value of the loan plus interest. This value may now be compared with the value of the debts.

The quotient of the expectation value and of the fully discounted nominal value of the loan

demand is here no different fromthe survival chance of the loan taking into account breakdown

distributions.

χ

∗

=

D

L ·

1 +i

1 +i

s

= 1 −ρ

∗

(7.7)

Shortfall Risk on Uncovered Loans: Option-Theory Approach 63

Equation (7.6) substituted into equation (7.7) results in:

1 −ρ

∗

=

L ·

1 +i

1 +i

s

− P

L ·

1 +i

1 +i

s

(7.8)

Further transformations lead to:

(1 −ρ

∗

) · L ·

1 +i

1 +i

s

= L ·

1 +i

1 +i

s

− P (7.9)

L ·

1 +i

1 +i

s

−ρ

∗

· L ·

1 +i

1 +i

s

= L ·

1 +i

1 +i

s

− P (7.10)

ρ

∗

· L ·

1 +i

1 +i

s

= P (7.11)

ρ

∗

=

P

L ·

1 +i

1 +i

s

(7.12)

Equations (7.11) and (7.12) give in detail the correlation between the put value and the credit

shortfall risk probability taking breakdown distributions into account.

7.3 DERIVATION OF RISK-ADJUSTED VALUES

First the assumption is made of a company that has its debts position only in the form of a

bank loan. If it is intended that the loan interest rate i be risk adjusted, then the value of the

debts corresponds, in the case of only one bank loan, to its nominal value:

D = L (7.13)

By substituting equation (7.6) into equation (7.13) one obtains:

L = L ·

1 +i

1 +i

s

− P (7.14)

Now comes the decisive difference from the approaches to date made by Black/Scholes and

Merton. With the aid of equation (4.49), which demonstrates the relation between the risk-free

standard rate of interest i

s

, the shortfall risk ρ

∗

and the risk-adjusted rate of interest i , the

quotient in the brackets of equation (7.14) may be remodelled.

1 +i

1 +i

s

=

1 +

i

s

+ρ

∗

1 −ρ

∗

1 +i

s

=

1 −ρ

∗

+i

s

+ρ

∗

(1 +i

s

) · (1 −ρ

∗

)

=

1 +i

s

(1 +i

s

) · (1 −ρ

∗

)

(7.15)

The result reads:

1 +i

1 +i

s

=

1

1 −ρ

∗

(7.16)

64 Risk-adjusted Lending Conditions

Substituted into equation (7.14) this gives:

L =

L

(1 −ρ

∗

)

− P (7.17)

Elimination from equation (7.17) using ρ

∗

leads to:

ρ

∗

=

P

L + P

(7.18)

From this it follows that the credit shortfall risk for the calculation of the risk-adjusted loan

interest rate is known, if one succeeds in calculating the value of the put.

According to equation (7.18) it follows from P = 0 that ρ

∗

= 0, and for P = ∞that ρ

∗

= 1.

All values that P may assume thus result in a permissible value for ρ

∗

, and vice versa.

To test this, the credit shortfall risk starting from equation (7.12) will be calculated once

more. Equation (7.12) is, with the help of equation (7.16) expressed as follows:

ρ

∗

=

P

L ·

1

1 −ρ

∗

(7.19)

Extended by (1 −ρ

∗

) results in:

ρ

∗

=

P · (1 −ρ

∗

)

L

(7.20)

Further transformations result in:

ρ

∗

· L = P −ρ

∗

· P (7.21)

ρ

∗

· (L + P) = P (7.22)

ρ

∗

=

P

L + P

(7.23)

The comparison shows that equations (7.18) and (7.23) are identical.

What is essentially new in the approach presented here is that the relation between i

s

, i and

ρ

∗

according to equation (7.16) is used, ﬁrst to calculate the credit shortfall risk ρ

∗

, detached

from the risk-free standard rate of interest i

s

and from the loan interest rate i . This is the small

detour, already mentioned in Chapter 1, which does, however, lead in what follows to the

general solution.

If χ

∗

= 1 −ρ

∗

(see equation (2.8)) is calculated ﬁrst, the later calculations can be simpliﬁed.

Under equations (7.18) and (7.23) it is the case that:

1 −χ

∗

=

P

L + P

(7.24)

Transformations result in:

L + P −χ

∗

(L + P) = P (7.25)

χ

∗

· (L + P) = L (7.26)

χ

∗

=

L

L + P

(7.27)

Shortfall Risk on Uncovered Loans: Option-Theory Approach 65

According to Black/Scholes, for a period of time following on [CORU85, S. 211], the value

of P amounts to:

P = L ·

(1 +i )

(1 +i

s

)

· N(x) − V · N(x −σ) (7.28)

with: x =

ln

L · (1 +i )

V · (1 +i

s

)

σ

+

σ

2

(7.29)

N = standard normal distribution function

σ = volatility of the company value according to the time period

On this we make the assumption here that the alternative investment might be made at the

i

s

rate of interest. This is the equivalent of granting credit to a borrower that would be, for

the bank, risk-free. As such borrowers do not in reality exist this assumption might appear

fanciful. This is not, however, the case, as the following consideration indicates: the alternative

investment is made again at the risk-adjusted rate of interest to a borrower that is real and

does exist. If the rate of interest is in fact established as risk adjusted, the bank’s yield on this

investment is, after deduction of losses, again equivalent to the rate of interest i

s

. This means,

in other words, that a bank’s yield in any loan transaction always corresponds to the risk-free

standard rate of interest i

s

, providing the loan interest rate i is in fact calculated so that it is

risk adjusted. The i

s

may thus take over the function of the risk-free rate of interest in terms

of being a model in the context of the lending operations of any bank.

With the deﬁnition of debt rate

d =

L

V

(7.30)

d = debt rate

equation (7.27) expresses itself, by applying equations (7.28), (7.16) and (2.8) abbreviated to

V, as follows:

χ

∗

=

d

d +

d

χ

∗

· N(x) − N(x −σ)

(7.31)

Multiplying out results in:

χ

∗

· (d − N(x −σ)) +d · N(x) = d (7.32)

Reverse substitution χ

∗

= 1 −ρ

∗

results in:

(1 −ρ

∗

) · (d − N(x −σ)) = d · (1 − N(x)) (7.33)

ρ

∗

· (d − N(x −σ)) = d − N(x −σ) −d +d · N(x) (7.34)

ρ

∗

=

d · N(x) − N(x −σ)

d − N(x −σ)

(7.35)

with: x =

ln

d

1 −ρ

∗

σ

+

σ

2

(7.36)

66 Risk-adjusted Lending Conditions

As ρ

∗

also occurs in the expression for x, the set of equations (7.35)–(7.36) must be solved

iteratively. This presents no problemwith today’s standard PCsoftware (for example Microsoft

Excel Release 4 or higher, see Appendix 2).

The approximation [CORU85, S. 205] N(x) ≈ N · (x −s) ≈ 1 applies where d/

(1 −ρ

∗

) 1. It follows from this that ρ

∗

= 1. This means that ρ

∗

approaches 1 for high

degrees of outside indebtedness.

In the reverse case, where d/(1 −ρ

∗

) 1, the approximation N(x) ≈ N · (x −s) ≈ 0

applies. It follows from this that ρ

∗

≈ 0. This means if the outside indebtedness approaches

zero, the credit shortfall risk is also approaching zero.

Highly interesting is the fact that the credit shortfall risk ρ

∗

is now only dependent on the

degree of outside indebtedness d and on the volatility σ of the company’s market value. As

the volatilityis, however, dependent onthe time periodand/or loantermunder consideration, the

duration of these also has matching inﬂuence. In particular the credit shortfall risk is, however,

not dependent on the risk-free standard rate of interest i

s

! This is the essential difference from

the Black/Scholes result using the zero bond approach [CORU85, S. 382].

The term of the loan that has been newly paid out, and/or the remaining term of a loan that

it is intended to reassess, may be chosen as the period of time. Volatility relating to one year

will be assumed for σ, as it is annual accounts and budgets that are normally evaluated in the

context of loan assessment. The solution thus runs as follows:

ρ

∗

(t ) =

N(x −σ ·

√

t ) −d · N(x)

N(x −σ ·

√

t ) −d

(7.37)

with: x =

ln

d

1 −ρ

∗

(t )

σ ·

√

t

+

σ ·

√

t

2

(7.38)

t = loan/term

It is important to note here that ρ

∗

(t ) is calculated with equations (7.37–7.38). ρ

∗

(t ) is thus

the probability of credit shortfall over the whole loan term. This value has to be converted to

the mean credit shortfall risk per annum, aided by the rules for transforming notice periods

(Section 3.2), in order to be able to calculate the annual rate of interest.

ρ

∗

pa

= 1 −

t

1 −ρ

∗

(t ) (7.39)

The annual rate of interest consistent with risk can then be calculated with the aid of

equation (4.49):

i

pa

i

spa

+ρ

∗

pa

1 −ρ

∗

pa

(7.40)

i

pa

= interest rate on annual basis

i

spa

= risk-free interest rate on annual basis

ρ

∗

pa

= average annual credit shortfall risk

The annual volatility should be inserted here for σ, and the term in years for t . If various

new loans with different terms are granted to a company, then the appropriate credit shortfall

risk must be decided for each term and proceeded with as described above.

Shortfall Risk on Uncovered Loans: Option-Theory Approach 67

Inthe case of current account loans, the time betweentwoassessment dates shouldbe inserted

for t , i.e. normally one year. It follows from this, however, that the submission of intermediate

accounts has an improving effect on credit-worthiness, as the bank’s time for reaction becomes

shorter, and a correspondingly shorter period may be inserted into the formula for t . It lines up

with the current banking practice of demanding intermediate accounts of borrowers at higher

risk. As has been shown, this makes complete sense.

It is still being assumed, on the basis of the approach used here, that loan interest is only

paid for the whole term of the loan when the loan is repayable. This is of course not always

the case! What does this now imply for the application of the method described here? Even if

interest payments become payable during the term of a loan, it may happen that the borrower

is not in a position to pay them. Owing to his inability to make interest payments, the borrower

may thus already go bankrupt, before the loan itself is in default. This is not taken into account

here. That, by implication, means the following:

r

Either, that the probability that the borrower already goes bankrupt owing to a loan interest

payment becoming due, is so small that it may be ignored,

r

Or, that the borrower has been given time to make the interest payments, possibly until the

repayment of the loan is due.

Only empirical investigations may show whether or not these implicit assumptions are

permissible. At this point the supposition may merely be expressed that they do apply; it may

in all probability be assumed that this ﬁrst assumption applies in the case of ﬁrms of good

ﬁnancial standing. The second assumption will, however, always apply in the case of ﬁrms of

poor ﬁnancial standing, if the bank extending the credit is justiﬁed in hoping for a recovery in

the situation. If there is no such hope, then it makes no difference if the borrower goes bankrupt

earlier or only when repayment of the loan becomes due. If the put remains in the money in any

case, it does not have to be waited for. Compare this to Brealey and Meyers [BRMY96, S. 668].

In Section 11.3 we demonstrate, with the help of a real-life example, how interest payments

and loan repayments may be taken into account prior to expiry of the loan term.

7.4 DETERMINATION OF THE VALUES FOR

THE SOLUTION FORMULA

This chapter is concerned with giving the user of our method useful hints on how it may

be applied in practice. For this it is assumed that the borrower’s track record and budget

documentation are available to the user. First we explain how to proceed when the borrower is

a company. The analogous conclusion of how to proceed for a private borrower is put forward

in a ﬁnal subsection.

7.4.1 The Value of the Company and its Debt Rate

The expositions in this subsection are based on explanations by Brealey and Meyers

[BRMY96]. The debt-free company serves as a starting point for our considerations. In this

case the market value of the assets corresponds to the market value of the equity, as the owners

of the equity do not have to fulﬁl any expectations for providers of outside capital.

68 Risk-adjusted Lending Conditions

According to Brealey and Meyers, the value of a company may be determined by discounting

its free cash ﬂows and totalling the resulting values [BRMY96, S. 71/71]. The free cash ﬂow

is here deﬁned as revenue less costs less investments [BRMY96, S. 71]:

Free cash ﬂow = Revenues − Costs before interest and taxation − Investments

It is important to emphasise at this point that the free cash ﬂowis calculated with costs included

prior to interest and taxation. We are indeed concerned here at ﬁrst to determine the value

of the assets alone, without bring the liabilities side of the company’s balance sheet into the

picture at all!

In the notation being used in this study, the value of companies is therefore calculated as

follows [BRMY96, S. 72]:

V =

∞

¸

n=1

C

(1 +i

d

)

n

(7.41)

C = free cash ﬂow

i

d

= discount rate

The total according to equation (7.41) is calculated as follows [BRMY96, S. 49]:

V =

C

i

d

(7.42)

(but also compare equation (7.46) and the statements made there).

The discount rate used according to equation (7.42) is, along with free cash ﬂow, of decisive

importance. It would, however, go far beyond the scope of this study to go any more closely

into the determination of the discount rate at this juncture. Sufﬁce is to point out here that in

today’s literature, despite some shortcomings that are widely recognised, the CAPM (capital

asset pricing model) is as favoured as ever:

i

d

= i

g

+β(i

mt

−i

g

) (7.43)

β = measurement of unleveraged assets market risk according to CAPM

i

g

= return on a risk-free investment in government bonds

i

mt

= return on the market

We refer you to the literature concerned for further expositions on the CAPM, for example

[BRMY96, S. 143–236].

The debt rate, which comes into place for the method described here, is therefore calculated

as follows:

d =

L

V

=

i

d

· L

C

(7.44)

Summarising, let it be emphasised that it is a question, in the case of C, of the average,

future free cash ﬂow of the company.

An importance consequence ensues from equation (7.44): the debt rate relevant to lending

is independent of the debt rate used for accounting. Only the discounted future free cash ﬂows

count. This may mean, in an extreme case, that a company that is today overloaded with debt

Shortfall Risk on Uncovered Loans: Option-Theory Approach 69

according to its books may still be credit-worthy, provided its future prospects are otherwise

good. The method being described does therefore provide the answer in relation to the extent

to which it may be worth restructuring a company that is, according to its books, overloaded

with debt.

Recently Cho [CHO98] has shown, with the aid of empirical investigations, that the structure

of the company’s ownership does, when all is said and done, have inﬂuence on the company’s

value.

This study shows once more how complex is the question of valuing companies. Whether

the very simple procedure we sketch out is sufﬁcient to do justice to the method described

here, only empirical enquiries will show.

7.4.2 Volatility

In order to gain some idea of the volatility of a company it may, on the one hand, be ascertained

on the basis of historical data. Here again, however, it is the case that the future is no simple

extrapolation of the past. Better to combine the two, using annual accounts going back three

or four years and one to two future-orientated budgets, even though these are associated with

uncertainties. On a purely theoretical basis alone, however, it is not possible to lay down how

volatility should be determined correctly for the purposes of the model. This can only be

determined with the help of empirical investigations (see Chapter 10).

Substantially better results in determining volatility could be achieved if borrowers produced

accounts every six months or even quarterly. But this is often not the case, as considerable

work would be involved and no one apart from banks would be asking for them. As long as

there are banks that do not insist on intermediate accounts this is not going to gain acceptance

in the market. If such intermediate accounts did exist, substantially longer series of facts would

be available for recently observed periods of time and results would improve accordingly. In

so far as intermediate accounts do exist, it is therefore recommended that they be used.

The concrete calculation of annual volatility takes place by analogy with the method that

was described by Cox and Rubinstein [CORU85, S. 254 and thereafter]:

σ =

Γ

n −1

2

Γ

n

2

·

1

2

·

n

¸

k=1

ln

V

k+1

V

k

−µ

2

(7.45)

µ =

1

n

·

n

¸

k=1

ln

V

k+1

V

k

**n = number of quotients of annual reports and budgets
**

Γ = gamma function

µ = medium of the logarithms

The gamma function values are of interest as shown in Table 7.1 [KREY91, S. 159].

The correction factor in front of the root in equation (7.45) for various values of n is shown

in Table 7.2.

70 Risk-adjusted Lending Conditions

Table 7.1 Gamma function values

x 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Γ(x)

√

π 1

√

π

2

1

3 ·

√

π

4

2

15 ·

√

π

8

6

105 ·

√

π

16

24

1.7725 1.0000 0.8862 1.0000 1.3293 2.0000 3.3234 6.0000 11.631 24.000

Table 7.2 Correction factor values

n 2 3 4 5 6 7 8 9 10

Γ

n −1

2

Γ(

n

2

)

1.7725 1.1284 0.8862 0.7523 0.6647 0.6018 0.5539 0.5158 0.4847

The values of a company for individual ﬁnancial years (historical annual accounts and

budgets) should likewise be calculated according to equation (7.42). The expositions given in

the previous subsection apply regarding this. In relation to earlier annual accounts, we must —

at this point —go particularly into years with negative free cash ﬂows: it makes no sense to

discount negative free cash ﬂows. The reﬂection that every company disposes of a liquidation

value, even if it may only be very small, is of further assistance here. From the point of view

of the lending bank, the value of a company is never lower than the liquidation value that can

be expected. This leads to the following equation for the value of a company, which must be

substituted into equation (7.42):

V

k

= Max

C

k

i

dk

; V

lk

(7.46)

If necessary, equation (7.46) for V should also be used in equation (7.44), especially in

cases of winding up and reconstruction. The credit-worthiness and ﬁnancial standing of any

company in fact depend, strictly speaking, more than anything on liquidation value (see also

Section 7.6).

The Black/Scholes model assumes volatility is constant. Merton [MERT73] showed later

that the Black/Scholes model may be applied as it were unchanged, if volatility is not constant

but is known as a function of time. The only difference is the deﬁnition of σ

2

as the average

over the remaining term T of the option:

σ

2

(T) = T

−1

·

T

0

σ

2

(t ) dt (7.47)

It is thus sufﬁcient, to be able to apply the method described here, to have an adequately

accurate idea of σ

2

over the loan’s term.

Determination of the relevant volatility plays a central part in the whole option theory.

It has only been possible here to show a practical method that is effective in application.

The difﬁculties in determining volatility can be seen in, for example [SWD192], [RSS093],

[KAR093], [HULL97, S. 499 and thereafter], [BCCH97] and [RITR98].

Shortfall Risk on Uncovered Loans: Option-Theory Approach 71

7.4.3 Private Debtors

The situation as regards private debtors can be shown by analogy with that of companies.

With private debtors free cash ﬂow is the equivalent of discretionary income after deduction

of essential living costs. The more luxuries come to be regarded as necessities the smaller the

borrowing power of the private individual becomes, and vice versa. For this reason, determining

the correct ﬁgure to put on discretionary income at any one time, by talking to each individual

borrower, is a tricky task for bankers.

Next we have to determine the appropriate discount rate i

d

, with which the quasi-market

value of the individual is found out. The CAPM may not be applied here, but the following

reﬂection helps further, namely that the capability, in the case of the individual, to make interest

payments and repayments, is of decisive importance. It is therefore appropriate to apply the

highest possible loan interest rate that could be expected, taking into consideration any possible

repayments, for the purposes of discounting. In the case of consumer credit, this is normally

the legal interest rate cap. If the method outlined here ends up with a loan interest rate that is

higher than the legal interest rate cap, then the business does not get done anyway.

By contrast with companies, two further risks have to be assessed in connection with an

individual’s discretionary income: the risks of death (and incapacity to work) and of unem-

ployment. The former risk may be covered by the insurance industry on attractive terms. So in

granting credit to individuals it is extremely important to take out appropriate insurance and

to ensure the proceeds of it are assigned in favour of the bank.

The risk that the borrower may become unemployed must be taken into account by an

adjustment to the ﬁgure for discretionary income. First the rate of unemployment beneﬁt

applicable, on the basis of the borrower’s personal data in respect of age, occupation and

residence, must be determined. Then the ﬁgure for discretionary income must be multiplied

by a factor of 1 less the above-determined unemployment beneﬁt, such that one obtains the

expectation value of the discretionary income taking the risk of unemployment into account.

Once discretionary income for a few years back has been determined in this way —and

used as a budget for the future —these values may be used instead of the free cash ﬂows in

the formulae described above. In this way loans to individuals also become assessable.

Only discretionary income has been taken into account in the reﬂections outlined here —any

assets that may exist have not been considered. As no asset should be taken into account, by

way of precaution, unless it is assigned in favour of the loan, the rules for covered loans apply,

and these will be described in Chapters 8 and 9.

7.5 INFLUENCE OF INDIVIDUAL PARAMETERS

ON THE CREDIT SHORTFALL RISK

As was shown in the previous section, the credit shortfall risk according to our model depends

on the following inﬂuence factors:

r

The company’s debt rate d

r

The annual volatility of the company’s value σ

r

The loan’s term in years t

72 Risk-adjusted Lending Conditions

0% 50% 100%

100%

50%

0%

d

Parameter:

0.5

0.3

0.2

t = 1

ρ*(d = 100%) = 100% ! !

Figure 7.5

It is intended to demonstrate the inﬂuence of these parameters on the credit shortfall risk

according to our model, with the help of the following diagrams.

Even though it has not been possible to date to carry out any costly empirical tests —for the

reason that no bank in Switzerland has the necessary historically documented data available —

the following diagrams do in fact line up qualitatively with the author’s professional experience

to date, which goes back at least 15 years.

Figure 7.5shows that anappropriate credit shortfall riskswitches inaccordingtothe volatility

of the company’s value, once a certain debt rate has been passed. It is important to note, at this

point, that the following always applies:

ρ

∗

(d = 100%) = 100%!

If the loan granted is higher than the company’s market value, then the risk exposure is 100%!

Figure 7.6 corresponds with Figure 7.5, but the credit shortfall risk axis has been spread

out. Here it is evident that low credit shortfall risks result, even in the case of high debt rates,

providing the volatility of the company’s value is correspondingly low.

Figure 7.7 corresponds with Figure 7.6 to the extent that the debt rate axis has been spread

out too. Here it is evident that lowcredit shortfall risks result, even in the case of very high debt

rates such as those of banks and insurance companies, provided the volatility of the company’s

value is very low.

Shortfall Risk on Uncovered Loans: Option-Theory Approach 73

0% 50% 100%

2%

1%

0% d

Parameter:

0.5

0.2

0.1

0.05

t = 1

Figure 7.6

90% 95% 100%

2%

1%

0%

d

Parameter:

0.05

0.02

0.01

0.1

t = 1

Figure 7.7

74 Risk-adjusted Lending Conditions

0 0.5 1

100%

50%

0%

Parameter: d

60%

40%

20%

80%

t = 1

Figure 7.8

Figure 7.8 corresponds to Figure 7.5, but σ and d have been exchanged. Here it is evident that

low credit shortfall risks result, even in the case of high volatility in company value, provided

the debt rate is correspondingly low. Comparison between Figures 7.5 and 7.8 suggests the

conclusion that more than anything else it is important for an entrepreneur to keep the volatility

of the market value of his company low, and not necessarily its debt rate (compare comments

in Section 9.4).

Figure 7.9 corresponds with Figure 7.8, with the credit shortfall risk axis again spread out.

Here too it is evident that low credit shortfall risks result even in the case of high debt rates,

provided the volatility of the value of the company is low.

Figure 7.10 demonstrates the course of the credit shortfall risk independently of the term of

the loan and of the volatility of the company’s value. The dependence of the credit-worthiness

of a loan on the term of the loan with the same borrower, is clearly discernible.

Figure 7.11 corresponds to Figure 7.10, but over shorter loan terms.

Figure 7.12 corresponds to Figure 7.10, where the volatility of the company’s value and its

debt rate have changed places.

Figure 7.13 corresponds to Figure 7.12, but over shorter loan terms.

Figure 7.14shows whichpairs of values (σ, d) matchthe same credit shortfall risk. According

to the guidelines of the rating agencies, a debtor receives a AAA/Aaa rating, if the shortfall risk

amounts to less than one millionth over the whole term. The line with ρ

∗

(t ) = 1 ppm therefore

represents the AAA curve to the extent that all pairs of values below this curve match this top

rating.

Shortfall Risk on Uncovered Loans: Option-Theory Approach 75

0 0.2 0.4

2%

1%

0%

Parameter: d

80%

60%

40%

t = 1

Figure 7.9

4%

2%

0%

0 1 5

t

Parameter:

0.5

0.2

0.1

2 3 4

1%

3%

5%

Years

d = 60%

Figure 7.10

76 Risk-adjusted Lending Conditions

0%

0 3

t

Parameter:

0.5

0.2

0.35

6 9 12

1%

2%

Months

d = 60%

Figure 7.11

4%

2%

0%

0 1 5

t

Parameter: d

80%

60%

40%

2 3 4

1%

3%

5%

Years

= 0.2

Figure 7.12

Shortfall Risk on Uncovered Loans: Option-Theory Approach 77

4%

2%

0%

0 3

t

Parameter: d

90%

80%

70%

6 9 12

1%

3%

5%

Months

= 0.2

Figure 7.13

100%

50%

0%

0% 25% 50% 75% 100%

1%

1bp

1ppm

25%

75%

d

Parameter: *(t )

Figure 7.14

78 Risk-adjusted Lending Conditions

7.6 RISK OF BANKRUPTCY AND BREAKDOWN DISTRIBUTION

Up till now it has been the credit shortfall risk ρ

∗

that has been calculated. At this juncture

we should demonstrate how the risk of a borrower’s bankruptcy and the probably breakdown

distributions may be calculated.

As may be inferred fromFigure 7.3 in Section 7.2, the probability that a borrower is bankrupt

at the time of the loan becoming due for repayment is none other than that of the put being

in the money. This probability can be determined according to Cox and Rubinstein, with the

notation introduced there [CORU85] being used in the paragraph printed in italics below, in

order to be able to understand the derivation better.

The probability that a call is in the money at maturity amounts to Φ[a; n, p] [CORU85,

S. 177], by using the complementary binomial distribution function. This probability receives

the value N(x −σ ·

√

t ) [CORU85, S. 208] in the case of transition to the standard normal

distribution function. The probability of the corresponding put at maturity being in the money

is therefore 1 − N(x −σ ·

√

t ) [CORU85, S. 4]. This becomes N(x +σ ·

√

t ) on the basis of

the symmetry properties of the standard normal distribution function [CORU85, S. 211].

Returning to the notation being used here, we can thus write, analogously, as follows:

ρ = N(x) (7.48)

x =

ln

d

1 −ρ

∗

σ ·

√

t

+

σ ·

√

t

2

as per (7.38)

With the credit shortfall risk ρ

∗

determined, the probability of bankruptcy ρ can thus also be

determined. The breakdown distribution rate probability is calculated, according to equation

(2.4) by

b = 1 −

ρ

∗

ρ

= 1 −

ρ

∗

N(x)

(7.49)

and the expected breakdown distributions by:

B = b · L · (1 +i (t )) =

1 −

ρ

∗

N(x)

· L · (1 +i (t )) =

1 −

ρ

∗

ρ

· L · (1 +i (t )) (7.50)

i (t ) = loan interest rate according to the whole term

As was emphasised at the end of Section 7.3, by way of qualiﬁcation, the calculations in

this chapter are based on the assumption that the loan interest for the whole term of the loan is

due for payment only when the loan itself is due to be repaid. In the same way the calculation

of the breakdown distribution rate is also based on this assumption, which should be taken into

account here. In other words, the breakdown distribution rate relates to the nominal value of

the loan plus all interest and compound interest over the whole term of the loan.

Let us remember, at this point, that the minimum market value of a company corresponds,

according to equation (7.46) to its liquidation value. This leads to the following reﬂections:

r

Companies of good ﬁnancial standing have high revenues and thus high market values,

which are always higher than the balance sheet ﬁgures produced by their accountants (and

let us remember at this point, too, that it is the market value of the debt-free company that is

Shortfall Risk on Uncovered Loans: Option-Theory Approach 79

used for risk calculation). Loans are thus small in size in relation to the company’s market

value. That leads to very low shortfall risks —in the main just theoretical.

r

On the other hand companies of poor ﬁnancial standing have low revenues and thus low

market values. As soon as a serious crisis looms, it may thus be assumed that it is the

liquidation value that is used, according to subsection 7.4.2, for the risk calculation, which

lines up with banking practice. This may lead consequentially to an existing loan in this

situation being called in, or to an application for a new loan not being followed up.

7.7 LOAN ASSESSMENT

The derivation of the risk-adjusted loan interest rate when extending credit having been anal-

ysed in Sections 7.3 to 7.6, this section will deal with assessing a loan that has already been

granted during its term, in which on the basis of the loan agreement the loan interest rate up

to the time the loan is due to be repaid can no longer be adjusted to be consistent with risk,

although new information for assessing it is available. The basic equations from Section 7.2

again form the starting point for our considerations.

The value of a loan during its term can be calculated with the aid of equations (7.6) and

(7.12), in which we must, however, distinguish between the market and the nominal value

of the loan. This leads to the following equation, in which (1 +i )/(1 +i

s

) = 1/(1 −ρ

∗

) no

longer applies, as the interest rate during the contractually agreed term of the loan may no

longer be adjusted to be consistent with risk:

Λ = L ·

1 +i

pa

1 +i

spa

t

− L ·

1 +i

pa

1 +i

spa

t

· N(x) + V · N(x −σ ·

√

t ) (7.51)

Summarised, this results in:

Λ = L ·

1 +i

pa

1 +i

spa

t

· (1 − N(x)) + V · N(x −σ ·

√

t ) (7.52)

with: x =

ln

d ·

1 +i

pa

1 +i

spa

t

σ ·

√

t

+

σ ·

√

t

2

(7.53)

In this, t means the rest of the loan’s termand the interest rates i

pa

and i

spa

and the volatility σ

must be inserted on an annual basis. Moreover, the current standard rate of interest appropriate

to the remainder of the term must be used for i

spa

. As will be evident from equations (7.48)–

(7.49), Λ = L, if i

pa

according to ρ

∗

is consistent with risk!

Unlike Section 7.3, where in the risk-adjusted case for d > 1 there is no longer any solution

for ρ

∗

(or where ρ

∗

= 100%) (cf. notes on Figure 7.5), there is a solution here for all values

of d. This is explained by the fact that it is indeed normal that conditions may no longer be

adjusted to be consistent with risk in new circumstances during the agreed ﬁxed term of the

loan. This gives expression to the fact that it is only the situation when the loan is due for

repayment that is decisive. Even if the value of d is greater than 1 during the term, the market

value of the company through to the end of the term may improve again, and the situation of d

being less than 1 may be achieved again by the time the loan is due for repayment. Expressed in

80 Risk-adjusted Lending Conditions

terms of option theory: even if the put during the loan’s term is occasionally in the money, that

does not necessarily mean that it is also in the money at maturity (compare also with Brealey

and Myers [BRMY96, S. 564]).

For the case in which the liabilities side is made up of several loans, the debts as a whole

must be inserted for L and assessed according to the remaining term of the loan in question.

The value of the loan being considered then works out in proportion to its share of the debts.

Here the values of the loan under consideration must be used for the rates of interest i

pa

and

i

spa

. Tackled this way we can be certain that the loan in question is assessed according to its

characteristics, but taking into account the total of debt (see also Section 8.4).

7.8 BONDS

Bonds can be assessed in exactly the same way as bank loans, according to Section 7.7. With

the help of equations (7.52)–(7.53), however, the converse, namely the volatility implicit in

the stock exchange, may be calculated on the basis of the stock exchange price.

With the aid of volatility investigated in this way, the shortfall risk implicit in the stock

exchange, and therefore the bond’s rating, may be determined using equations (7.37)–(7.38).

7.9 CONSIDERATION OF PRIVILEGED SALARY CLAIMS

IN THE EVENT OF BANKRUPTCY

The salary claims of a company’s workforce are privileged in the case of bankruptcy, under

Swiss debt recovery and bankruptcy law. Under this the date on which bankruptcy proceedings

are commenced is taken as the date on which notice of termination of employment is given. The

privileged demand of each person employed is that person’s salary claim up until the expiry of

the notice period according to his/her contract of employment. In contrast to all other relevant

liabilities in bankruptcy, the company does not have to account for these salary claims prior to

bankruptcy. These salary claims have thus not been taken into account in the considerations in

this chapter so far. This shall now be remedied.

Under equation (7.50) the breakdown distribution probability value comes to:

B = b · L · (1 +i (t )) =

1 −

ρ

∗

ρ

· L · (1 +i (t )) (7.50)

First of all the privileged salary claims have now to be deducted from the expectation value

of the breakdown distribution available for distribution:

B

c

= b · L · (1 +i (t )) − S =

1 −

ρ

∗

ρ

· L · (1 +i (t )) − S (7.54)

B

c

= corrected breakdown distribution

S = proportional salaries

It should be noted that, in the case of several loans, only the proportional salaries have to be

taken into account for any individual loan that has to be assessed. This implies the following: in

bankruptcy the whole amount for salaries is deducted from the overall breakdown distribution.

Shortfall Risk on Uncovered Loans: Option-Theory Approach 81

The rest is distributed proportionately across the individual loan demands. From the point

of view of the individual lender, however, this means nothing more than that it has to make

a proportionate contribution to the privileged salary claims that have to be met, from the

breakdown distributions that were ‘originally’ available to it.

The percentage rate of the salary total falling against it in this way corresponds to percentage

ﬁgure of its loan claim in relation to the total loan.

It is quite possible that the salary claims in bankruptcy are higher than the probable break-

down distributions. Thus a negative ﬁgure for corrected probable breakdown distributions

would result, which would make no sense. This just means that the salary claims are no longer

covered at all. Equation (7.50) must therefore be correctly written as follows:

B

c

= Max(b · L · (1 +i (t )) − S; 0) (7.55)

Continuing, the corrected breakdown distribution rate probability can be detailed as follows:

b

c

=

B

c

L · (1 +i

c

(t ))

=

Max(b · L · (1 +i (t )) − S; 0)

L · (1 +i

c

(t ))

(7.56)

b

c

= corrected breakdown distribution rate

i

c

(t ) = corrected loan interest rate according to the whole term

It is essential at this point to note the distinction between i (t ) and i

c

(t )! i (t ) concerns the

uncorrected rate of interest that has already been calculated, while i

c

(t ) concerns the corrected

value still to be calculated.

The corrected credit risk can now be calculated with the aid of equation (2.6):

ρ

∗

c

= ρ · (1 −b

c

) = ρ ·

1 −

Max(b · L · (1 +i (t )); 0)

L · (1 +i

c

(t ))

(7.57)

ρ

∗

c

= corrected credit risk

For the case where b

c

= 0, the solution is:

ρ

∗

c

= ρ if b

c

= 0 (7.58)

For other cases the following applies:

ρ

∗

c

= ρ · (1 −b

c

) = ρ ·

1 −

B

c

L · (1 +i

c

(t ))

(7.59)

On the other hand, the corrected credit risk may be calculated by:

i

c

(t ) =

i

s

+ρ

∗

c

1 −ρ

∗

c

=

i

s

+ρ · (1 −b

c

)

1 −ρ · (1 −b

c

)

(7.60)

Both equations (7.59) and (7.60) now permit calculation of the corrected breakdown distri-

bution rate probability b

c

. The middle and right-hand term of equation (7.59) are used. Cancel

out with ρ, subtract from 1 and cancel out by −1 results in:

b

c

=

B

c

L · (1 +i

c

(t ))

=

B

c

L ·

1 +

i

s

(t ) +ρ · (1 −b

c

)

1 −ρ · (1 −b

c

)

(7.61)

82 Risk-adjusted Lending Conditions

Transformation of the denominator in the right-hand term of equation (7.61) gives:

b

c

=

B

c

L ·

(1 +i

s

(t ))

(1 −ρ · (1 −b

c

))

=

B

c

· (1 −ρ · (1 −b

c

))

L · (1 +i

s

(t ))

(7.62)

Reduce to the lowest common denominator using the denominator in the right-hand term of

equation (7.62) and multiply out gives:

b

c

· L · (1 +i

s

) = B

c

−ρ · B

c

+ρ · B

c

· b

c

(7.63)

Solving by b

c

leads to:

b

c

=

B

c

· (1 −ρ)

L · (1 +i

s

(t )) − B

c

· ρ

(7.64)

The corrected credit shortfall risk is thus:

ρ

∗

c

= ρ · (1 −b

c

) = ρ ·

1 −

B

c

· (1 −ρ)

L · (1 +i

s

(t )) − B

c

· ρ

(7.65)

This credit has still to be converted on an annual basis according to equation (7.39).

7.10 LIMITS TO THE APPLICATION OF THE OPTION

THEORY APPROACH

The preconditions that have to be fulﬁlled in order to be able to apply the Black and Scholes

model are speciﬁed in Cox and Rubinstein [CORU85, S. 268]. They are explained below, with

the terms having been ‘translated’ for the application described here.

1. The company does not make any distributionpayments (dividends andinterest) during

the period under consideration.

The method described here takes the free cash ﬂows of the debt-free imaginary company

as the basis for assessing its credit-worthiness. The assessment rests thus on the amount of

revenue before interest and dividends. In the assessment this ﬁgure is determined for each

ﬁnancial year being taken into consideration, independently of the other ﬁnancial years.

The distributions of dividends and payment of interest thus do not have any inﬂuence on

the assessment in the method described here, as long as the approved and calculated loans

are not changed (raised) as a result of the payment of dividends and interest.

2. Notice to terminate the loan may only be given at the end of the period under consid-

eration.

This precondition is always fulﬁlled.

3. There are no requirements in terms of margins, taxation or transaction expenses.

This precondition is fulﬁlled as it is just a question, in the case of the put in the application

described here, of a theoretical construction and not of anything existing in real life.

4. The level of interest is constant.

As was shown in this chapter, the credit-worthiness of a company is on the one hand

independent of the interest level; in contrast to the solution as it was indicated in Cox and

Shortfall Risk on Uncovered Loans: Option-Theory Approach 83

Rubinstein [CORU85, S. 382] (see Section 7.1). The assessment of the market risk on the

other hand was intentionally excluded here (see Section 1.2).

5. The volatility of the company’s value is constant.

The approach described here may only be applied if reliable assumptions are made about

future volatility. Only the calculating out of various scenarios can show whether or not the

uncertainties are decisive in terms of price. In so far as this is the case, the appropriate

consequences have to be drawn; either the price has to be set in line with the worst-case

scenario, or no loan must be made at all. The precondition of constant volatility was qualiﬁed

by Merton [MERT 73], as we have already mentioned: compare also subsection 7.4.2 with

the reading references given there. It must not, however, be denied that volatility represents

the crux of all applications of option price models. Even Cox and Rubinstein write [in

CORU85, S. 258] that ‘the critical feature for optionpricingis the behaviour of the volatility’.

6. Only very small changes in the valuation of a company can occur in very short periods

of time.

This precondition may usually be regarded as fulﬁlled. Major changes are often attributable

to errors in earlier assessments and are therefore artiﬁcial. In any such case the errors should

be corrected and the assessment undertaken afresh.

A further precondition is mentioned in Cox and Rubinstein [CORU85, S. 276].

7. Changes in the value of a company must be distributed lognormally.

There are statistical tests to check whether any existing frequency distribution out of n

observations (here: appraised company accounts and budgets), that has been obtained as a

result of a randomsample, is consistent with a hypothesis that has been drawn on distribution

in the parent population (here: lognormal distribution), and these tests are described in the

reading (for example, BOHL92, S. 625 and thereafter). Because of the statistically small

volumes of data in the form of appraised company accounts and budgets, use of the visual

test by means of a probability paper (for example, BOHL92, S. 625 and thereafter) is

recommended —it being possible these days to run this up on to a screen with the help of

the computer [BOHL92, S. 629].

The further back a company’s track record goes, the more precisely can it normally be

established whether or not the condition is fulﬁlled. One qualiﬁcation results from the point

that track records may only be used as far back as there may not have occurred, in the period

concerned, any substantial events changing the nature of the company. By this we mean

events such as mergers, divestments, major changes in commercial strategies and so on.

Kremer and Roenfeldt [KRRO92] moreover pose the question of whether the Black and

Scholes model is usable at all for longer-term examinations. They argue that ofﬁcially

quoted options have a maximum term of 270 days. The Black/Scholes assumption within

this time scale, in which the value of the asset base is changing continually but only in

small steps, may well be permissible, but not over longer examination time scales. In

their article about warrants, they conclude that their ‘jump-diffusion’ model in the case

of warrants exerciseable in more than one year (for out of money warrants) may deliver

more reliable results than the Black and Scholes model. Translated into extending credit,

this means a loan assessment endures only so long as the company laying claim to it

is not subject to any substantial changes of the kind mentioned above. The requirement

that loan agreement clauses must contain appropriate passages follows on from this. Even

84 Risk-adjusted Lending Conditions

for long-term loans, such as ﬁxed-date mortgage loans over several years, it must thus

be possible, if important events deﬁned in advance (see above) occur, to adjust the loan

conditions to new circumstances.

A further qualiﬁcation emerges on the point of the extent to which information exists at

all. A company must after all have a ‘track record’ in the ﬁrst place, for an assessment of it

to be possible. This, however, is of course not the case with recently founded enterprises.

The only way forward in this situation is to work on the basis of budgets. Here the risk must

be checked, with the aid of model calculations and the help of specimen scenarios, to see

whether it can be circumscribed with sufﬁcient precision, or the company applying for the

loan must be prepared to pay interest on a worst-case scenario basis.

7.11 RESULTS AND CONCLUSIONS

A company’s credit shortfall risk can be calculated using equations (7.37)–(7.38), indepen-

dently of the risk-free rate of interest:

ρ

∗

(t ) =

N(x −σ ·

√

t ) −d · N(x)

N(x −σ ·

√

t ) −d

(7.37)

with: x =

ln

d

1 −ρ

∗

(t )

σ ·

√

t

+

σ ·

√

t

2

(7.38)

As ρ

∗

(t ) in the above equations corresponds to the credit shortfall risk over the whole term of

the loan, the value should be calculated on an annual basis:

ρ

∗

pa

= 1 −

t

1 −ρ

∗

(t ) (7.39)

The risk-adjusted loan interest rate per annum is then calculated from that, as follows:

i

pa

=

i

spa

+ρ

∗

pa

1 −ρ

∗

pa

(7.40)

A loan that has already been granted may be continually assessed for risk-adjustment

right through to its maturity. Assessment is made in the case of t years according to the

equations:

Λ(t ) = L ·

1 +i

pa

1 +i

spa

t

· (1 − N(x)) + V · N(x −σ ·

√

t ) (7.52)

with: x =

ln

d ·

1 +i

pa

1 +i

spa

t

σ

√

t

+

σ ·

√

t

2

(7.53)

The risk-adjusted loan interest rate can also take privileged salary claims into account in

the event of bankruptcy: ﬁrst the corrected expectation value of breakdown distributions is

Shortfall Risk on Uncovered Loans: Option-Theory Approach 85

calculated:

B

c

= Max · (b · L · (1 +i (t )) − S; 0) (7.55)

The corrected credit shortfall risk may then be calculated:

ρ

∗

c

= ρ ·

1 −

B

c

· (1 −ρ)

L · (1 +i

s

(t )) − B

c

· ρ

(7.65)

This value must then be converted again according to equation (7.39). The loan interest rate

also ensues from equation (7.40). An Excel worksheet is presented in Appendix 2 with which

the equations speciﬁed above may be applied.

7.12 EXAMPLES

The method outlined here is explained with the aid of two examples.

7.12.1 Example of a Company with Continuous Business Development

The key ﬁgures needed for the model are given in Table 7.3.

The liquidation value is lower in all years than the discounted free cash ﬂows, which is why

this applies as the company’s value, according to equation (7.46).

Debts are summarised together as follows:

Creditors 50

1-year loan(s) 500

3-year loan(s) 1000

Total debts 1550

Debt rate: 1550/2500 = 62%

The natural logarithms must ﬁrst be calculated according to equation (7.42) for the calcula-

tion of volatility:

Table 7.3 Key ﬁgures —Example 1

+1

Year −3 −2 −1 0 (Budget)

Turnover 1000 1025 1100 1150 1250

Operating costs 700 750 800 850 900

Capital investments 100 110 100 90 100

Free cash ﬂow 200 165 200 210 250

Discount rate 10% 10% 10% 10% 10%

Discounted free cash ﬂows 2000 1650 2000 2100 2500

Liquidation value 1000 1000 1000 1000 1000

Value of company 2000 1650 2000 2100 2500

86 Risk-adjusted Lending Conditions

Table 7.4 Shortfall risks in Example 1

σ

√

t x N(x) N(x −σ

√

t ) ρ

∗

1 year 19.25% −2.3844 0.0086 0.0050 0.0518%

3 years 33.34% −1.2195 0.1113 0.0602 1.5724%

Table 7.5 Final results of Example 1

ρ

∗

pa

according i

s

i according to

Rating to rating (assumption) equation (4.49) i rounded up

1 year AA 0.0733% 4.0% 4.0763% 4

1

8

%

3 years BB 0.7570% 4.5% 5.2971% 5

5

16

%

Table 7.6 Bankruptcy situation in Example 1

ρ

∗

ρ b L(1 +i (t )) B

1 year 0.0518% 0.8554% 93.94% 520 489

3 years 1.5724% 11.1329% 85.88% 1159 996

ln(1650/2000) = −0.1924

ln(2000/1650) = 0.1924

ln(2100/2000) = 0.0488

ln(2500/2100) = 0.1744

mean µ = 0.0558

Using equation (7.45) a volatility of σ = 19.25% results from this.

The iterative calculation delivers the results shown in Table 7.4, using an Excel worksheet.

The result for ρ

∗

must now be converted to one year, according to the rules for transforming

time periods (equation (7.39)):

1 year: ρ

∗

pa

= 0.0518%

3 years: ρ

∗

pa

= 0.5269%

The one-year loan receives an AA rating and the three-year loan receives a BB rating

according to Table 2.1. This leads to the ﬁnal results for the loan rates of interest consistent

with risk, according to Table 7.5. It is striking that despite the relatively small volatility of not

quite 20% and a debt rate of almost two-thirds, there is a considerable difference in interest

between the one-year and the three-year loan.

The situation that may nowbe expected in any possible case of bankruptcy may be calculated,

using equations (7.48, 7.49 and 7.50) and shown in Table 7.6.

Shortfall Risk on Uncovered Loans: Option-Theory Approach 87

Table 7.7 Proportion of privileged salary claims

Liabilities Amount Share of salary claims

Creditors 50 2

Loan 1 Year 500 20

Loan 3 Years 1000 40

Total 1550 62

Table 7.8 Final results after privileged salary claims

i

s

i according i

B

c

b

c

(%) ρ

∗

c

(%) ρ

∗

cpa

(%) Rating (assumption) to (4.49) rounded up

1 year 469 90.12 0.849 0.849 A 4.0% 4.1780% 4

3

16

%

3 years 956 82.07 1.9959 0.6698 BB 4.5% 5.2971% 5

5

16

%

If the entrepreneur gives up his company after one year, the bank may thus assume that the

loans, including accumulated interest over the whole terms of them, still have a value of about

94% and after three years a value of about 86%. (Here the interest must be calculated precisely

using the values of ρ

∗

given in Table 7.4.)

It is now our intention to investigate, in the event of bankruptcy according to Section 7.9,

what effect the salary claims have on the credit risk. The salary and wage claims to be expected

in the event of bankruptcy are 62. They spread out over the liabilities, as in Section 7.9, as

shown in Table 7.7.

This leads, using equations (7.55), (7.64) and (7.65) together with the values fromTable 7.7,

to the corrected credit shortfall risks, ratings and loan interest rates shown in Table 7.8.

Taking the salary claims in the event of bankruptcy into account in the case of the one-year

loan has led to deterioration in the credit-worthiness by one rating level. The rating level has

not deteriorated in the case of the three-year loan, despite an increase in risk. This comes about

because the ‘higher’ rating levels are ‘broader’ than the ‘lower’ rating levels, and because the

values in the case concerned are correspondingly favourable.

7.12.2 Example of a Company with a Poor Financial Year

The key ﬁgures needed are again given in Table 7.9.

In year –1 the liquidation value is higher than the discounted free cash ﬂows, which is why

this value is put in as the company’s value. Debts are summarised together as follows:

Creditors 50

1-year loan(s) 500

3-year loan(s) 500

Total debts 1050

88 Risk-adjusted Lending Conditions

Table 7.9 Starting position Example 2

+1

Year −3 −2 −1 0 (Budget)

Turnover 1000 1025 1100 1150 1200

Operating costs 700 750 800 850 900

Capital investments 100 120 50 100 100

Free cash ﬂow 200 230 50 100 200

Discount rate 10% 10% 10% 10% 10%

Discounted free cash ﬂows 2000 2300 500 1000 2000

Liquidation value 1000 1000 1000 1000 1000

Value of company 2000 2300 1000 1100 2000

Table 7.10 Shortfall risks in Example 2

σ

√

t x N(x) N · (x −σ

√

t ) ρ

∗

1 year 68.47% −0.4307 0.3333 0.1323 10.8658%

3 years 118.59% 0.5070 0.6939 0.2486 41.8652%

Debt rate: 1050/2500 = 52.5%

The volatility calculation is made analogously to the previous example:

ln(2300/2000) = 0.1398

ln(1000/2300) = −0.8329

ln(1000/1000) = 0.0000

ln(2000/1000) = 0.6931

mean µ = 0.0000

That results in a volatility of 68.47% according to equation (7.45). The iterative calculation

delivers the results shown in Table 7.10, using an Excel worksheet.

The transformation of time periods according to equation (7.39) gives the following results:

1 year: ρ

∗

pa

= 10.87%

3 years: ρ

∗

pa

= 16.54%

Using Table 2.1, this leads to the ratings and ﬁnal results for the loan rates of interest consistent

with risk shown in Table 7.11.

The poor ﬁnancial year − 1 has increased the volatility of the company’s value markedly.

The loan rates of interest consistent with risk are substantially higher than in the previous

example, despite the clearly lower debt.

Shortfall Risk on Uncovered Loans: Option-Theory Approach 89

Table 7.11 Final results of Example 2

ρ

∗

pa

according i

s

i according to

Rating to rating (assumption) equation (4.49) i rounded up

1 year C 12.4786% 4.0% 18.8281% 18

7

8

%

3 years DDD 24.9871% 4.5% 39.2993% 39

5

16

%

Table 7.12 Bankruptcy situation in Example 2

ρ

∗

ρ b L(1 +i (t )) B

1 year 10.87% 33.33% 67.40% 583 393

3 years 41.87% 69.39% 39.67% 981 389

The situation shown in Table 7.12 is to be expected in any possible event of bankruptcy.

Taking the salary claims in the event of bankruptcy into account by analogy with the previous

example is left to the reader.

Comparison of the results of the two examples in subsections 7.12.1 and 7.12.2 permits the

supposition that in today’s banking practice, in cases of companies having had poor ﬁnancial

years in the recent past and whose values are therefore highly volatile, loan interest rates have

been allowed to apply that are much too low.

8

Loans Covered against Shortfall Risk

In this chapter we will examine the shortfall risk of covered loans, where the bank in principle

relies exclusively on the collateral. As was illustrated in the comments on Figure 7.5, no loan

may ever be greater, at the time of its being issued, than the market value of the company or

the equivalent of that in respect of the private individual. The same applies, analogously, for

loans that are covered, but on which the bank relies on the collateral alone: the loan may not

be higher than the value of the collateral, so long as it is geared to the collateral alone. The

smaller the loan in relation to the value of the collateral, the smaller is the risk on the loan, and

vice versa. This fact is probed in Section 8.1.

In Sections 8.2 and 8.3 we will, in passing, relax this assumption, in that in the case of the

collateral falling short the loan will not necessarily be in default as long as the borrower is still

in a position to service it. This way of looking at it rests on the assumption that banks usually

only make loans when borrowers are, to all intents and purposes, in a position to service them

without any difﬁculty. Collateral here only serves the bank as guarantee against the case that

is out of the ordinary. In Section 8.2 the correlation between the probabilities of the collateral

falling short and of the borrower defaulting will ﬁrst be examined. The results of Sections 8.1

and 8.2 will be summarised in Section 8.3.

How to proceed in the case of a combination of one covered loan and one uncovered loan

to the same borrower will be investigated in Section 8.4. Here only the fact that one loan is

covered, and the other not, will be taken into account, with the consequences which ensue from

that fact. The question of the optimum combination of covered and uncovered loans, and thus

also the question of a loan that is partially covered, will not be investigated until Chapter 9.

The results of this chapter will be illustrated in Section 8.6 by means of an example.

8.1 SHORTFALL RISK OF A COVERED LOAN ON THE BASIS

OF THE OPTION-THEORY APPROACH

A covered loan, where the bank is relying exclusively on the cover, is in default if the value

of the collateral no longer matches the mortgaging bank’s minimum claim. This consideration

leads to the same result for the company as in Chapter 7, with terms being substituted as shown

in Table 8.1.

In the case of covered loans the time from the completion of the loan to the realisation of

the collateral must also be taken into account in the bank’s reaction time. Experience shows

that this can take from just a few days in the case of securities offered as collateral, to several

years in the case of mortgages.

In the case of securities it is possible to reassess the collateral daily, and electronically, on

the basis of stock exchange data. More information is thus available for the calculation of

92 Risk-adjusted Lending Conditions

Table 8.1 Uncovered versus covered loans

Variable Chapter 7 Chapter 8

V Market value of the company Value of collateral

D, L Debts Mortgage

E Equity Unmortgaged portion of collateral

σ Volatility of the value of the company Volatility of the value of the collateral

d Debt rate Mortgage rate

t (Remaining)term of loan and/or reaction

time in the case of current account lending

(Remaining)term of loan and/or reaction

time in the case of current account lend-

ing and variable mortgages

volatility in the case of securities used as collateral, which leads to a simpliﬁcation vis-` a-vis

equation (7.27) [CORU85, S. 256]:

σ =

A

n −1

·

n

¸

k=1

ln

X

k

X

k

−1

−µ

2

(8.1)

with µ =

1

n

·

n

¸

k=1

ln

X

k

X

k

−1

X = portfolio values

A = number of trading days per year

In order to be able to assess mortgage loans, the current value of the property concerned

must be known, together with the volatility of this value. The determination of the current

value of a property is common practice for any bank. It cannot actually be said that this is free

of problems, but they are known and largely under control.

It is more difﬁcult with the determination of volatility. For this not only the current value of

any property but also the probable development in its value has to be known. Credit is due to

the Cantonal Bank of Z¨ urich’s pioneering work in this ﬁeld, whereby it developed a property

price index based on the ‘hedonistic’ method. This method not only allows for the published

index to be drawn up, but also makes it possible for an individual calculation of the probable

development in the value of any property to be drawn up. See [ZKB96] for details. Index

valuations may be downloaded from the Internet (see Appendix 3).

The loan risk of the covered loan is calculated using equations (7.37)–(7.38), the shortfall

risk of the collateral using equation (7.48) and the breakdown distribution probability value

using equation (7.49). It is important to distinguish clearly between these values. They are

therefore detailed once more at this juncture:

p

∗

C

=

N

C

(x

C

−σ

C

·

√

t ) −d

C

· N

C

(x

C

)

N

C

(x

C

−σ

C

·

√

t ) −d

C

(7.37)

Loans Covered against Shortfall Risk 93

with x =

ln

d

1−ρ

∗

σ ·

√

t

+

σ ·

√

t

2

(7.38)

ρ

C

= N

C

(x

C

) (7.47)

b

C

= 1 −

ρ

∗

C

ρ

C

=

ρ

∗

C

N

C

(x

C

)

(7.48)

Index C = collateral

8.2 CORRELATION BETWEEN THE SHORTFALL RISK

OF THE BORROWER AND THE SHORTFALL RISK

OF THE COLLATERAL

If a covered loan defaults only on account of the collateral, that does not necessarily mean

that the loan is effectively lost to the bank. The borrower may indeed be in as good a position

as previously to meet its obligations to the bank on the strength of its solvency. It is still of

course just as much as before the bank’s contractual partner. This section will therefore deal

with determining the combined shortfall risk of the borrower and of the collateral, taking the

correlation between the two into account.

8.2.1 Derivation of the Correlation

The probabilities of four possible occurrences within a period of time have to be considered

in the case of a covered loan:

r

Probability ˆ a, that both the borrower defaults and the collateral falls short.

r

Probability

ˆ

b, that neither the borrower defaults nor the collateral falls short.

r

Probability ˆ c, that only the collateral falls short, but that the borrower does not default.

r

Probability

ˆ

d, that the borrower defaults, but that the collateral does not fall short.

This can be illustrated graphically as shown in Figure 8.1.

d b

a c

0

0

1

1

collateral

borrower

The ‘default’ occurrence has a value of ‘1’.

The ‘no default’ occurrence has a value of ‘0’.

Figure 8.1

94 Risk-adjusted Lending Conditions

The following correlations are discernible on the basis of Figure 8.1.

ρ

B ∩C

= ˆ a

ρ

B

= ˆ a +

ˆ

d

ρ

C

= ˆ a + ˆ c

1 = ˆ a +

ˆ

b + ˆ c +

ˆ

d

(8.2)

ρ

B ∩C

= combined risk

Index B = borrower

One obtains the following [KREY91, S. 304] for the correlation of this frequency

distribution:

ˆ r =

ˆ a −( ˆ a +

ˆ

d) · ( ˆ a + ˆ c)

[( ˆ a +

ˆ

d) −( ˆ a +

ˆ

d)

2

] · [( ˆ a + ˆ c) −( ˆ a + ˆ c)

2

]

(8.3)

ˆ r = correlation coefﬁcient

After insertion of the shortfall risks according to equation (8.2), one obtains:

ˆ r =

ρ

B ∩C

−ρ

B

· ρ

C

¸

ρ

B

−ρ

2

B

¸

·

¸

ρ

C

−ρ

2

C

¸

(8.4)

Solution using ρ

B ∩C

results in:

ρ

B ∩C

= ρ

B

· ρ

C

+ ˆ r ·

ρ

B

−ρ

2

B

·

ρ

C

−ρ

2

C

(8.5)

It is then possible to calculate the combined shortfall risk, provided the individual shortfall

risks and the correlation coefﬁcient are known. In the special case ˆ r = 0, i.e. in which there

is no correlation at all, equation (8.5) reduces itself to ρ

B ∩C

= ρ

B

· ρ

C

, as one would expect

according to the multiplication rules that apply to the probability of independent occurrences

[BOHL92, S. 324].

8.2.2 Value Area of the Efﬁciency of the Correlation

It is intended next to investigate what values may theoretically be assumed for ˆ r and ρ

B ∩C

.

Minimum

First of all the minimum should be calculated. The frequency distribution may then be por-

trayed as follows and shown graphically in Figure 8.2, whereby this involves a theoretically

conceivable perfect hedge that is, however, unrealistic in practice.

ρ

B ∩C

= 0

ρ

B

=

ˆ

d

ρ

C

= ˆ c

1 =

ˆ

b + ˆ c +

ˆ

d

(8.6)

Loans Covered against Shortfall Risk 95

c

b d

collateral

borrower

1

1

0

0

0

The ‘default’ occurrence has a value of ‘1’.

The ‘no default’ occurrence has a value of ‘0’.

Figure 8.2

One obtains the following [KREY91, S. 304] for the correlation of this frequency

distribution:

ˆ r =

0 −

ˆ

d · ˆ c

(

ˆ

d −

ˆ

d

2

) · (ˆ c − ˆ c

2

)

(8.7)

After insertion of the shortfall risks according to equation (8.6), one obtains:

ˆ r

min

=

−ρ

B

· ρ

C

ρ

B

−ρ

2

B

·

ρ

C

−ρ

2

C

(8.8)

According to statistical theory, the lowest value that a correlation coefﬁcient may assume is

−1 [BOHL92, S.235]. We now intend to investigate under what assumptions this value may

be reached.

−1 =

−ρ

B

· ρ

C

ρ

B

−ρ

2

B

·

ρ

C

−ρ

2

C

(8.9)

Conversion results in:

ρ

B

−ρ

2

B

·

ρ

C

−ρ

2

C

= ρ

2

B

· ρ

2

C

(8.10)

Multiplying out gives:

ρ

B

· ρ

C

−

ρ

B

· ρ

2

C

+ρ

C

· ρ

2

B

+ρ

2

B

· ρ

2

C

= ρ

2

B

· ρ

2

C

(8.11)

Which leads to:

1 = ρ

B

+ρ

C

(8.12)

ˆ r may thus only assume the value −1 if the sum of both individual risks is equal to 1. This

means, on the basis of equation (8.6), that

ˆ

b = 0 in this case.

Maximum

Maximum values for ˆ r and ρ

B ∩C

are reached by the consideration that such is the case if

collateral falling short leads to default on the loan, i.e. ρ

B ∩C

= ρ

C

. The frequency distribution

96 Risk-adjusted Lending Conditions

a

d b

1

1

0

0

0

collateral

borrower

The ‘default’ occurrence has a value of ‘1’.

The ‘no default’ occurrence has a value of ‘0’.

Figure 8.3

can then be portrayed as follows and illustrated in Figure 8.3.

ρ

B ∩C

= ˆ a

ρ

B

= ˆ a +

ˆ

d

ρ

C

= ˆ a

1 = ˆ a +

ˆ

b +

ˆ

d

(8.13)

As

ˆ

d ³ 0 applies as probability value for

ˆ

d, this means that ρ

B

> ρ

C

. The shortfall risk

of the borrower is thus greater or at least equal to the shortfall risk of the collateral and to

the combined shortfall risk respectively. This makes sense, as the bank would of course not

otherwise fall back on collateral at the time of granting the loan. One obtains the following

[KREY91, S. 304] for the correlation of this frequency distribution:

ˆ r

max

=

ˆ a −( ˆ a +

ˆ

d) · ˆ a

[( ˆ a +

ˆ

d) −( ˆ a +

ˆ

d)

2

] · [ ˆ a − ˆ a

2

]

(8.14)

After insertion of the shortfall risks according to equation (8.12), one obtains:

ˆ r

max

=

ρ

C

−ρ

B

· ρ

C

ρ

B

−ρ

2

B

·

ρ

C

−ρ

2

C

(8.15)

According to statistical theory, the highest value that a correlation coefﬁcient may assume

is +1 [BOHL92, S.235]. We intend therefore to investigate under what assumptions this value

may be reached.

1 =

ρ

C

−ρ

B

· ρ

C

ρ

B

−ρ

2

B

·

ρ

C

−ρ

2

C

(8.16)

Conversion results in:

ρ

B

−ρ

2

B

·

ρ

C

−ρ

2

C

= (ρ

C

−ρ

B

· ρ

C

)

2

(8.17)

Multiplying out gives:

ρ

B

· ρ

C

−ρ

B

· ρ

2

C

−ρ

C

· ρ

2

B

+ρ

2

B

· ρ

2

C

= ρ

2

C

−2 · ρ

B

· ρ

2

C

+ρ

2

B

· ρ

2

C

(8.18)

Loans Covered against Shortfall Risk 97

Which leads to:

ρ

B

−ρ

2

B

−ρ

C

+ρ

B

· ρ

C

= 0 (8.19)

Taking out of brackets and abbreviation gives:

ρ

B

= ρ

C

(8.20)

ˆ r may thus only assume the value +1 if both individual risks are equally large. This means,

on the basis of equation (8.13), that

ˆ

d = 0. Borrower default and collateral fall short therefore

always occur either simultaneously or not at all.

8.3 SHORTFALL RISK OF THE COVERED LOAN

This section is concerned with bringing together the individual elements that have been calcu-

lated to date. Equation (8.5) details the risk that both the borrower defaults and the collateral

falls short, which is indeed a prerequisite for default occurring on a covered loan. This proba-

bility must be multiplied by the risk of loss (1 −b

C

), in order to obtain the shortfall risk of the

covered loan. In proceeding thus it is implicitly assumed that only the realisation of the col-

lateral makes breakdown distributions possible, but not the realisation of the borrower’s other

asset values. This does, however, make sense and is in line with the principle of conservatism.

(ρ

B ∩C

)

∗

= (ρ

B ∩C

) · (1 −b

C

) (8.21)

By insertion of equations (8.5) and (7.41) one obtains:

(ρ

B ∩C

)

∗

=

ρ

B

· ρ

C

+ ˆ r ·

ρ

B

−ρ

2

B

·

ρ

C

−ρ

2

C

·

ρ

∗

C

ρ

C

(8.22)

In the special case ˆ r = 0, i.e. in which there is no correlation at all, equation (8.20) reduces

itself to:

(ρ

B ∩C

)

∗

= ρ

B

· ρ

∗

C

if ˆ r = 0 (8.23)

In the special case of the maximumcorrelation, equation (8.22) reduces itself, after insertion

of equation (8.15) to:

(ρ

B ∩C

)

∗

= ρ

∗

C

if ˆ r = ˆ r

max

(8.24)

ρ

B

must be determined according to the rules fromSection 7.4 for applying equations (8.22)

and (8.23).

In practice the challenge consists above all in determining the correlation coefﬁcient ˆ r for

the various loan transactions empirically, using statistical methods. Here one may put forward

the supposition that the correlation coefﬁcient for many loan transactions lies either close to

zero (for instance, in the case of ﬁnancing owner-occupied houses) or close to the maximum

(for example, in the case of loans secured against collateral in the form of securities), and the

application of equations (8.23) and (8.24) is therefore permissible.

In the case of a correlation close to zero it may thus well occur that the value of the collateral

falls below the nominal amount of the loan, but that the borrower continues to meet his

obligations. This was often the case in the ﬁnancing of owner-occupied houses in the middle

98 Risk-adjusted Lending Conditions

of the 1990s: the value of the owner-occupied house had fallen sharply in the course of the

general crisis in the property market, but the borrower continued to have the same income as

he had previously. It is important in such a situation that bank does not lose its nerve and call

in the loan unnecessarily. It is better to proﬁt from the low correlation by requesting moderate

additional repayments (i.e. ones that are affordable by the borrower), until the value of the

collateral has returned to stand at a ‘reasonable’ level in relation to the amount of the loan.

In the case of loans secured against collateral in the form of securities, experience is quite

different. If the value of the portfolio of securities that has been mortgaged falls below

the loan’s nominal value, only a few borrowers are in a position to continue to service the

loan out of other income. The correlation coefﬁcient in such cases usually lies close to the

maximum.

The suppositions put forward here are indeed plausible, but have yet to be corroborated

empirically.

8.4 COVERED AND UNCOVERED LOANS

TO THE SAME BORROWER

Both covered and uncovered loans are regularly granted to companies simultaneously. The

question thus arises when assessing the individual loans of how to analyse the facts about

the company. The determination of the company’s shortfall risk itself is made using equation

(7.48). Portraying bankruptcy proceedings in the form of a model forms another step in the

assessment of the loans. Let us explain below what is meant by this.

First of all, the simplest case is assumed, in which a company’s debt consists merely of one

covered and one uncovered loan. It is simpler here to assess the covered loan. Because it has

preferential status when it comes to bankruptcy, equation (8.22) may be applied directly.

In assessing the uncovered loan the procedure has to be analogous to that described in

Section 7.9. First of all the breakdown distribution probability value has to be calculated in

relation to the whole amount of loan:

B =

1 −

ρ

∗

(L)

ρ(L)

· L · (1 +i

tot

) cf. (7.50)

in which L corresponds here to the sum of the covered and uncovered loans.

The claim that is covered must then be deducted from this total breakdown distribution

probability value. This action is consistent with the principle of conservatism. The bank that

grants the uncovered loan must of course assume, in the worst-case scenario as far as it is

concerned, that the preferential demand will be met in full. On the basis of the breakdown

distribution probability value of the uncovered loan calculated in this way, the credit risk in

relation to the loan’s nominal value may be calculated according to equation (7.49).

This procedure will be illustrated by means of an example in the next section but one. We

have to proceed analogously in the case of debts of complicated structures. The principle is

always the same: the breakdown distribution probability values have to be determined ﬁrst in

total, followed by scheduling allocation of the breakdown distributions according to the Swiss

laws on the recovery of debt and on bankruptcy.

Loans Covered against Shortfall Risk 99

8.5 RESULTS AND CONCLUSIONS

Loans where the bankrelies solelyonthe collateral maybe calculatedusingthe same calculation

rules as uncovered loans to companies. The necessary substitutions have to be made for this,

as listed in Section 8.1.

Just the calculation of the volatility of securities accounts is calculated according to a slightly

modiﬁed equation (8.1).

Even if a loan’s collateral does fall short, this does not necessarily mean that there will be

default on the loan. The borrower may nonetheless be in a position to service it. The shortfall

risk is calculated in this case as follows:

(ρ

B ∩C

)

∗

=

ρ

B

· ρ

C

+ ˆ r ·

ρ

B

−ρ

2

B

·

ρ

C

−ρ

2

C

·

ρ

∗

C

ρ

C

(8.22)

As we have demonstrated, it is plausible to assume that the correlation coefﬁcient ˆ r is very

small in the case of mortgages and, in contrast, very high in the case of loans with securities

as collateral. This consideration leads to the following, simpler expressions:

(ρ

B ∩C

)

∗

= ρ

B

· ρ

∗

C

if ˆ r = 0 (8.23)

(ρ

B ∩C

)

∗

= ρ

∗

C

if ˆ r = ˆ r

max

(8.24)

Here ˆ r

max

is not necessarily equal to +1! That is only the case if ρ

B

= ρ

C

.

Combinations of covered and uncovered loans to the same borrower may be calculated. The

procedure is described in the preceding section. We shall work through an example in the next

section.

If, in real life, the shortfall risk of the collateral is greater than the shortfall risk of the

borrower, then a partially covered loan should be calculated: the combination, therefore, of a

covered and an uncovered loan.

This will be elaborated in Chapter 9, in fact in Section 9.6.

8.6 EXAMPLE

The same data is applied for this example as in subsection 7.12.1. The single difference consists

in it involving a three-year ﬁxed mortgage on a loan of three years. The allocation of breakdown

distributions under the Swiss law on debt recovery and bankruptcy is:

1. mortgage

2. privileged wages and salary claims

3. uncovered loans

The correlation between the company’s shortfall risk and the shortfall risk of the collateral

may be set at zero.

A mortgage shortfall risk of 0.1% has been calculated according to equations (7.37)–(7.38)

in Section 8.1.

The shortfall risk of the mortgage combined with the shortfall risk of the company is ﬁrst

calculated on the basis of this starting position. According to Table 7.6 the shortfall risk of

100 Risk-adjusted Lending Conditions

the company over three years is ρ

B

= 11.13%. ˆ r continues to be equal to zero, according to

equation (8.23):

(ρ

B ∩C

)

∗

= 0.1113 · 0.001 = 0.011113%

This corresponds to an AAA rating as per Table 2.1 with a rating risk of 0.0244%. If the

risk-free rate of interest i

s

= 4.5% (Table 7.5), this results —using equation (4.49) —in a

mortgage rate of interest of 4.5255%, or rounded up to 4

9

16

%.

The assessment of the uncovered loan takes place in a second step. The probable breakdown

distribution rate expected in one year comes to 93.94% according to Table 7.6. Related to the

total debt of 1550 plus accumulated interest of 4.08% (Table 7.5), this yields a breakdown

distribution probability value of 1515. The preferential mortgage demand, including the two

years of interest to be expected of 4

9

16

% of 1093, together with the privileged wages and

salary claims of 62, have to be deducted from this value. This results in probable breakdown

distributions of 360 for the non-preferential claims. This breakdown distribution is now allo-

cated proportionately to the uncovered creditors of 50 and to the uncovered loan of 500 plus

accumulated interest.

The value of 360 for B

C

may be inserted into equations (7.61–7.65) for further calculations.

This produces a corrected breakdown distribution rate of 62.74% in the case of uncovered

claims of 550. (For the sake of simplicity, the creditors are also considered as claims on which

interest may be claimed. This is indeed not quite correct, but the error arising is small and leads

to a slightly increased risk. This is acceptable in the light of the principle of conservatism.)

One then obtains a corrected loan risk of 0.2648%. This is in line with a BBBrating. The rating

risk is 0.3663% according to Table 2.1. If the risk-free rate of interest i

s

is 4% (Table 7.5), a

loan interest rate of 4.3824%, or rounded up to 4

7

16

%, ensues using equation (4.49).

Comparison of the results of subsection 7.12.1 and this section (Table 8.2) gives a picture

of the loan interest rate and the rating.

The company was paying annual interest of a total of 74 prior to depositing collateral,

falling to 67 following doing so. The rating of the three-year loan has improved markedly,

as one would expect. Conversely, however, the rating of the one-year loan has deteriorated

markedly. This is on balance better both for the borrower and the bank: the borrower is paying

loan interest reduced by 7 and the bank has lower loan risk on account of the collateral, which

in turn takes the form of a lower total interest expectation. The lower risk for the bank ensues

above all from the fact that the breakdown distribution probability values for the creditors are

smaller than they were previously, following the deposit of collateral. Because of the collateral,

the bank will receive a higher proportion of the breakdown distribution probability values, the

Table 8.2 Final results

Term 3-year loan without collateral 3-year loan with collateral

1 year AA 4

1

8

% BBB 4

7

16

%

3 years BB 5

5

16

% AAA 4

9

16

%

Loans Covered against Shortfall Risk 101

total of which will remain the same. The depositing of collateral is, as one might expect, at the

expense of the creditors.

It may be assumed, additionally, that two different banks grant the two loans. After the

depositing of collateral in favour of one bank, the situation will deteriorate considerably for the

other, unless it is immediately informed of the new situation and can adjust its own conditions

directly to the new circumstances. The example shows clearly that in the case of borrowers

that seek loans from several banks, it is absolutely essential to have, in so far as uncovered

loans are granted, a restrictive clause on mortgaging in the loan agreement.

9

Calculation of the Combination of Loans

with the Lowest Interest Costs

In Chapter 8 we have shown how one must proceed if a company is simultaneously seeking

coveredanduncoveredloans. This chapter will be concernedwithlayingdownthat combination

of different loans which gives rise to the lowest possible loan costs for any company.

In determining such ﬁnance for a company, the marginal interest rate of a loan is of central

importance, which is why this subject is dealt with in Section 9.1 at the beginning of this

chapter. Building up to the cases of two (Section 9.2) and three (Section 9.3) loans, the rules

are derived for whatever number of loans one likes (Section 9.4) that lead to the ﬁnancing

that is the most favourable in terms of interest costs. We will demonstrate in Section 9.5 how

partially covered loans may be calculated, and in Section 9.6, at what ratio of debt to equity

the highest returns on equity may be achieved.

A central theme in the granting of loans is the acceptability of the debt servicing and,

associated with that, the maximum degree of debt. This theme will be covered in Section 9.7

on the strength of our ﬁndings regarding the most favourable ﬁnancing in terms of interest costs.

The results and conclusions will be summarised in Section 9.8, and illustrated in Section 9.9

with the aid of an example.

9.1 MARGINAL INTEREST RATE

As can be seen in the ﬁgures in Section 7.5, any increase in the debt rate or mortgaging has the

effect of an increase in the loan interest rate for the whole loan. The additional interest costs

do, however, only arise on account of the increase itself and thus correspond in principle to the

interest costs of the increase.

i

m

is deﬁned as the marginal interest rate, as the following shows:

i

m

= lim

L→0

i (L +L) · (L +L) −i (L) · L

L

(9.1)

The marginal rate of interest thus corresponds to that rate of interest that is received, if one

allocates the entirety of additional interest costs arising from an inﬁnitesimally small increase

in loan to that increase. This can be portrayed graphically as shown in Figure 9.1.

Equation (9.1) turns, as a result of multiplying out, into:

i

m

= lim

L→0

i (L +L) · L +i (L +L) · L −i (L) · L

L

(9.2)

104 Risk-adjusted Lending Conditions

L

i

i(L + ∆L)

L + ∆L L

i(L )

i

s

Figure 9.1

Transpositions yield:

i

m

= lim

L→0

i (L +L) · L

L

+ L · lim

L→0

i (L +L) −i (L)

L

(9.3)

The solution of the ﬁrst limit is insigniﬁcant, in that it is shortened by L. In the case of

the second limit, it is a question of a differential quotient. The solution, according to that, runs

as follows:

i

m

= i (L) + L ·

∂

∂L

i (L) (9.4)

The marginal rate of interest of a loan amounting to L is thus the sum of the rate of interest

of this loan together with the partial differential of the rate of interest, derived from the amount

of the loan and multiplied by the amount of the loan. The problememerges here of determining

this partial differential. To anticipate the reader’s question immediately —it is not possible to

do so algebraically! If equation (4.49) is inserted into equation (9.4), the credit shortfall risk in

equation (9.4) is preserved. As explained in Section 7.3, the credit shortfall risk has to be solved

iteratively. Thus there is no other solution than to solve this differential iteratively too. There

is, however, a simple possibility of determining the marginal interest rate approximately, in

that in addition to the rate of interest i (L), the two rates of interest i (L +L) and i (L −L)

are calculated (see Figure 9.2).

In principle this involves determining the gradient of the interest curve at point A. As can be

seen in the ﬁgure, the gradient of the straight lines through points AB is less than the gradient

of the interest curve at point A. On the other hand the gradient of the straight lines through

points AC is greater than the gradient of the interest curve at point A. The gradient of the

Combination of Loans 105

L

L L − ∆L L + ∆L

i

B

A

C

i(L + ∆L)

i(L − ∆L)

i(L )

i

s

Figure 9.2

interest curve at point A thus lies between the gradients of the two straight lines described.

The same applies for the gradient of the straight lines through points BC. The gradient of the

straight lines though points BC may thus be selected as an approximate value for the gradient

of the interest curve at point A.

The maximum absolute error that is made here corresponds to the difference in the slope of

the straight lines AC and BC, which can be calculated. The value of L must thus be chosen

so that this difference is sufﬁciently small for the corresponding application of the result. The

approximate value for the marginal rate of interest thus works out as:

i

m

= i (L) + L ·

i (L +L) −i (L −L)

2 · L

(9.5)

Figure 9.3 shows the correlation between interest rate i and marginal interest rate i

m

. It

is striking that i and i

m

may be approximately identical for a large ﬁeld of values of d. The

rise of i

m

is, however, from a speciﬁed value of d onward, substantially steeper than that

of i .

9.2 TWO LOANS

Every borrower’s aimis to have the lowest possible interest costs. Let us assume that a borrower

needs outside capital to the extent of L. This sum may be divided into two individual loans L

1

and L

2

(for example, one covered and one uncovered), such that L = L

1

+ L

2

, which means

that the following applies:

i

1

· L

1

+i

2

· L

2

= min with L = L

1

+ L

2

(9.6)

106 Risk-adjusted Lending Conditions

i

i

m

i

d

i

s

= 3%, = 0.2, t = 1

i

s

0%

0%

5%

10%

15%

20%

25%

50% 100%

Figure 9.3

By expressing L

2

as L − L

1

, the variable L

1

now has to be determined in such a way that

minimum interest costs arise. L

1

has to be chosen in such a way that the ﬁrst derivation from

equation (9.6) at L

1

results in zero:

∂

∂L

1

[i

1

· L

1

+(L − L

1

) · i

2

(L − L

1

)] = 0 (9.7)

According to the rules of differential calculus, one obtains:

i

1

(L

1

) + L

1

· i

1

(L

1

) −i

2

(L − L

1

) −i

2

(L − L

1

) = 0 (9.8)

Reverse substitution of L

2

= L − L

1

gives:

i

1

(L

1

) + L

1

· i

1

(L

1

) −i

2

(L

2

) − L

2

· i

2

(L

2

) = 0 (9.9)

But from equation (9.4) this is none other than:

i

m1

= i

m2

(9.10)

Thus the lowest interest costs arise when the overall amount of loan is so divided that the

two marginal interest rates are identical. That is plausible too, as otherwise part of any loan

could be replaced by making an increase in the other loan on more favourable conditions.

Combination of Loans 107

9.3 THREE LOANS

This is done in the same way as in the preceding section. Equation (9.8) is written here as

follows:

i

1

· L

1

+i

2

· L

2

+i

3

· L

3

= min with L = L

1

+ L

2

+ L

3

(9.11)

Now there are two variables, L

1

and L

2

. That means there are two partial differentials to be

formed following L

1

and L

2

, which must both be equal to zero. First the partial differential

following L

1

is formed:

∂

∂L

1

[i

1

· L

1

+i

2

· L

2

+(L − L

1

− L

2

) · i

3

(L − L

1

− L

2

)] = 0 (9.12)

According to the rules of differential calculus, and following reverse substitution of

(L − L

1

− L

2

) = L

3

, one obtains:

i

1

(L

1

) + L

1

· i

1

(L

1

) −i

3

(L

3

) − L

3

· i

3

(L

3

) = 0 (9.13)

This again gives:

i

m1

= i

m3

(9.14)

The partial differentiation following L

2

ensues, analogously:

i

m2

= i

m3

(9.15)

This leads to the ﬁnal result:

i

m1

= i

m2

= i

m3

(9.16)

In this case, too, the ﬁnancing that is the most favourable in terms of costs is attained when all

marginal interest rates are identical.

9.4 THE GENERAL CASE OF SEVERAL LOANS

This involves generalising the ﬁndings of the two preceding sections. We will thus investigate

the situation in which a borrower seeks n loans and wishes thereby to achieve interest costs

that are as low as possible. The following therefore applies:

n

¸

j =1

L

j

· i

j

= min with L

n

= L −

n−1

¸

j =1

L

j

(9.17)

There are now n −1 variables L

1

. . . L

n−1

. N −1 partial differentials have therefore to be

formed, and these may be written in the following way:

∂

∂L

k

·

¸

¸

¸

¸

L

k

· i

k

(L

k

) +

n−1

¸

j =1

j =k

L

j

· i

j

L

j

+

¸

¸

¸

L − L

k

−

n−1

¸

j =1

j =k

L

j

· i

n

¸

¸

¸

L − L

k

−

n−1

¸

j =1

j =k

L

j

= 0

(9.18)

108 Risk-adjusted Lending Conditions

According to the rules of differential calculus, one obtains:

i

k

(L

k

) + L

k

· i

k

(L

k

) −i

n

(L

n

) −i

n

(L

n

) = 0 (9.19)

for all k with 1 £ k £ n −1

But this is again none other than:

i

mk

= i

mn

for all k with 1 ≤ k ≤ n −1 (9.20)

So it also applies in the general case, that a company has the most favourable outside ﬁnance

in terms of interest costs when all marginal interest rates are identical.

It is worth reﬂecting at this point that the marginal interest rates are, at the end of the day,

dependent on the term of the loan. If loans with different terms are now optimised with each

other, then it must be taken into account that the optimisation may no longer necessarily be

attained at the time of the next extension of a loan, as the conditions of the other loans are

bound to have been ﬁrmly agreed. The optimum solution in such situations has to be found by

using model calculations on the basis of various scenarios.

9.5 PARTIALLY COVERED LOANS

It is frequently the case in banking practice that loans to companies are partially covered, i.e.

the value of the collateral is lower than the amount of the loan. This may make perfectly good

sense in certain situations, but we will not go into the reasons here.

A partially covered loan is, in the meaning of the theory described here, none other than the

combination of a covered and of an uncovered loan. First of all therefore the division between

the two part loans that is the most favourable in terms of interest costs must be determined

according to Section 9.2 for the calculation of this loan. Then each part loan must be assessed

according to Chapters 7 and 8, and the risk-adjusted rate of interest worked out. The rate of

interest for the loan as a whole is then worked out at the end as the weighted average of the

rates of interest for the two part loans, on the basis of the proportion that each bears to the total

loan.

9.6 MAXIMUM RETURN ON EQUITY

All entrepreneurs strive to achieve maximum return on equity invested. One means to that end

is the optimisation of the ratio between debt and equity. How to do that will be demonstrated

below, where it will be assumed that the long-term average return on assets is ﬁxed. There is

no inconsistency here, even if the return on overall assets is volatile in itself and in relation to

itself. As better years and worse years succeed each other, a more stable average may result.

The following new variables are introduced:

return on assets: v = EBIT/V

return rate: g = company proﬁt/V

return on equity: e = company proﬁt/E

Combination of Loans 109

On the basis of these deﬁnitions we can write:

g = v −i (d) · d (9.21)

e =

g

1 −d

=

v −i (d) · d

1 −d

(9.22)

In order to attain the maximum return on equity e, the debt rate d must selected in such a

way that the differential of equation (9.22) from d is equal to zero:

∂

∂d

¸

v −i (d) · d

1 −d

¸

= 0 (9.23)

According to the rules of differential calculus, one obtains:

−(i

(d) · d +i (d)) · (1 −d) +v −i (d) · d = 0 (9.24)

Solving by v one obtains:

v = i (d) · d +(i

(d) · d +i (d)) · (1 −d) (9.25)

This is, however:

v = i (d) · d +i

m

(d) · (1 −d) (9.26)

Equation (9.26) means precisely that any company has a debt/equity ratio that leads to a

maximum return on equity, if the debt rate is selected in such a way that the interest costs, plus

equity capital, multiplied by the marginal interest rate of the debt structure that has the most

favourable interest costs, corresponds to its EBIT. In individual cases this condition may again

only be solved iteratively.

On the basis of the derivation the above statement applies in principle only for one company

with one loan. The general case with n loans can be derived likewise. The return on equity

now amounts to:

e =

v −

n

¸

j =1

i (d

j

) · d

j

1 −

n

¸

j =1

d

j

(9.27)

Whereby the debt rate was analysed into the n parts of the individual loan d

1

to d

n

. Now n

partial differentials have to be formed. The ﬁrst runs as follows:

∂

∂d

1

¸

¸

¸

¸

¸

v −i

1

(d

1

) · d

1

−

n

¸

j =2

i

j

(d

j

) · d

j

1 −d

1

−

n

¸

j =2

d

j

= 0 (9.28)

According to the rules of differential calculus, one obtains:

−[i

1

d

1

· d

1

+i

1

(d

1

)] ·

1 −d

1

−

n

¸

j =2

d

j

+v −i

1

(d

1

) · d

1

−

n

¸

j =2

i

j

(d

j

) · d

j

= 0 (9.29)

110 Risk-adjusted Lending Conditions

d

e = i

m

i

i / d/(e = i

m

)

i

s

= 3%, v = 10%, t = 1, = 0.2

0.1 0 0.2 0.3 0.4 0.5

10%

5% 50%

100%

0%

Figure 9.4

Solving by v and abbreviated with d = d

1

+

n

¸

j =2

d

j

one obtains:

v = i

m1

· (1 −d) +

n

¸

j =1

i

j

(d

j

) · d

j

(9.30)

The ﬁrst summand is again the equity rate multiplied by the marginal interest rate, and the

second summand corresponds to the average rate of interest on debt. It was demonstrated in

the preceding section (9.4) that the debt is structured at its most favourable in terms of interest

costs, when marginal interest rates are identical for all loans. The marginal interest rate i

m1

in equation (9.30) may thus be replaced by the general marginal interest rate i

m

. The rate of

interest that provides ﬁnance at the most favourable rate in terms of interest costs, according

to the preceding section (9.4), must be inserted as the marginal rate of interest.

The maximum return on equity may thus be calculated as:

e =

v −

n

¸

j =1

i

j

(d

j

) · d

j

1 −

n

¸

j =1

d

j

(9.31)

Combination of Loans 111

i

m

e i

d

i

s

EBIT = 10%, i

s

= 3%, = 0.2, t = 1

0%

0%

10%

20%

30%

30.96%

50% 81.8% 100%

Figure 9.5

According to equation (9.30) this is:

e =

i

m

· (1 −d) +

n

¸

j =1

i

j

(d

j

) · d

j

−

n

¸

j =1

i

j

(d

j

) · d

j

(1 −d)

= i

m

(9.32)

i.e., the maximum achievable return on equity corresponds to the marginal interest rate of the

debt structure that is the most favourable in terms of interest costs in the case of any appropriate

ratio of debt to equity. Figures 9.4 and 9.5 elucidate this correlation for any company with one

loan.

It may be recalled at this juncture that the market value, and not the balance sheet value,

has been used for the value of companies in preceding chapters. The statements made in this

section in relation to equity and return on equity must therefore be related to the market value:

equity = market value less debt

Maximum return on equity has therefore been derived on equity as deﬁned above, and not

on equity as deﬁned by the company’s books of account.

Figure 9.4 shows that high returns on equity may be achieved when the volatility of the

value of the company is low, even if the debt rate selected is very high and even if the credit

shortfall risk and loan interest rate are correspondingly high. The graphs, however, also make

it clear that the most important objective for any company must be to attain the lowest possible

112 Risk-adjusted Lending Conditions

volatility in its market value, in order to achieve the highest possible return on equity with high

debt rates (cf. also comments on Figure 7.8).

Conclusions

The marginal interest rate i

m

is always larger or equal to the rate of interest i . For the condition

of equation (9.26) to be fulﬁlled, the rate of interest i must therefore be lower than the return

on assets, and the marginal interest rate and return on equity respectively higher than return

on assets.

High volatility in the company’s market value and high rates of debt mean high interest

rates, and vice versa. It follows from this that, in the case of return on assets being equal, any

company the market value of which has higher volatility has a lower debt rate leading to a

maximum return on equity, than a comparable company, and vice versa.

If a company is ﬁnanced by loans over different terms, then the capital structure that leads

to maximum return on equity should be investigated by model calculations, with the aid of

different scenarios (see preceding Section 9.4). Here due heed should be given to the notion

of safety.

It is discernible from the course of return on equity e in Figure 9.5 that it is dangerous

to ﬁnance a company in such a way that maximum return on equity is achieved. Even small

increases in the debt rate lead in this situation to sharply reduced returns on equity.

The quintessence of this section results in the following: on the assumption that all banks

come to apply risk-adjusted loan conditions as described in this study, it is absolutely essential

for any company to keep revenues —and thus its market value —at the highest possible level,

with as little volatility as possible, in order to achieve a high return on equity via a high level

of debt made possible thereby.

9.7 ACCEPTABILITY OF DEBT SERVICING

In the preceding chapters we have demonstrated how risk-adjusted loan interest rates may be

calculated. What has until now been left out of consideration is whether the interest rates so

calculated are in general acceptable for companies. It obviously makes no sense at all to ﬁnance

any company at a high debt rate, if the risk-adjusted debt servicing that results from it leads

straight to bankruptcy.

9.7.1 Acceptability of Interest Rates

In the preceding section we have demonstrated how the ﬁnancing structure of a company may

be determined with the maximum achievable return on equity. The most important conclusion

at that point is the following: the loan interest rates of a company ﬁnanced from the point of

view of maximum achievable return on equity are lower than its return on assets. We must be

clear here that the return on assets has been calculated on the company’s market value and not

on its assets as deﬁned in its books of account. If this condition is fulﬁlled, the acceptability

of the interest is automatic.

Combination of Loans 113

9.7.2 Acceptability of Repayment

When it comes to repayments, the question that arises ﬁrst and foremost is why repayments

should be demanded from the bank at all. It may be argued that so long as a company is paying

risk-adjusted interest, it makes no sense for a bank to demand repayments —by doing so it

would indeed be passing up, at least partially, good business.

The answer to this question is illustrated in Figure 9.5. In the example selected there,

the company achieves a maximum return on equity if the debt rate is about 82%, and a re-

turn on equity of about 31% results from that. The interest rate on the debt amounts at this

point to about 5%, according to the i -curve. It is evident from the course of the e-curve

that even very small increases in the debt rate lead to a rapid falling away of return on eq-

uity, because the bank has, as a countermove, to raise the risk-adjusted loan interest rate

sharply. In the case of a debt rate of 90%, the risk-adjusted loan interest rate is doubled

to almost 10% and the return on equity sinks by about two-thirds to the same ﬁgure, of

about 10%.

The consequence of the course of the e-curve is therefore that it is dangerous to ﬁnance a

company to the level that matches maximum return on equity. Just one small but sharp drop in

earnings and the ensuing drop in market value, and the increase in debt rate arising therefrom

leads immediately to ﬁnancing on the basis of sharply reduced return on equity, and may even

lead to a case for winding up. In view of the danger outlined above, the optimum debt rate in

the example illustrated in Figure 9.5 lies somewhere between 60% and 80%. The return on

equity then is always still between 20% and 30%.

The following is now relevant as far as repayments are concerned. Let us take for granted

that any good, i.e. proﬁtable, company wishes to expand and has to invest accordingly. It may

be instructed to make an increase in its debt rate, under which it might come close to the point

of being ﬁnanced at the level at which return on equity is maximised. In doing so it would,

however, run into the dangers outlined above. It makes sense in such situations for the bank

to agree on repayments that in turn lead the company away from the ﬁnancial point at which

such dangers arise.

So the question that comes up now is at what level repayments should be set by the bank,

such that they are acceptable to the company. The upper limit here is reached if the company’s

owners pass up dividend payouts completely. Then all of the free cash ﬂow remaining after

interest and taxation may be applied to repayments. On the other hand this then means also

that there is nothing left there at all.

The following may now be relevant, reverting to the example in Figure 9.5. Let us take for

granted that the debt rate has risen to 80% as a result of new investment. The return on equity

then amounts to about 30%, i.e. the proﬁt on equity of 20% of the company’s market value

amounts to about 6% of the company’s market value. It follows from this that the company

could actually reduce the debt rate from 80% to 74% within one year. If the owners of the

company are entitled to 10% dividend payout, then the repayment is reduced to 4% of the

company’s market value. Then debt goes down within three years, other things being equal,

to below 70% and is then probably back again in the area that is safe. (To be correct, in this

example taxation should also have been taken into account. This was waived for the sake of

simplicity.)

114 Risk-adjusted Lending Conditions

9.7.3 Maximum Debt

It has become clear from the expositions made in this section so far that no company should be

granted more loans than are appropriate to the ﬁnancing of it for maximum return on equity. It

is, however —as outlined —dangerous to approach this maximum too closely. Just how far a

bank wishes to go, and with what levels of repayment —and how quickly —it wishes to lead

its borrower back from this danger point is, at the end of the day, a business policy decision.

The economic environment does of course play a major part in this too.

9.7.4 Consequences for Companies with Declining Earnings

Any bank in the business of lending money must be prepared for companies seeking loans

having to accept sharp drops in earnings. Here we intend to examine the resulting consequences.

A sharp drop in earnings for any company that has been safely ﬁnanced as described above,

means —logically —that the following scenarios have occurred:

r

reduction in the company’s market value

r

increase in debt rate

r

possible increase in the volatility of the market value

r

increase in the risk-adjusted loan interest rate

r

the debt rate having been drawn close to that at which return on equity is maximised or

having exceeded it (see Figure 9.5).

The consequences resulting for the bank are:

r

increasing the loan interest rate

r

demanding repayments, in order to attain the required safety distance from the maximum

debt rate (see Figure 9.5).

Provided the sharp drop in earnings does not occur too suddenly —i.e. providing the nec-

essary repayments are still acceptable to the company (see subsection 9.7.2) —there is no risk

for the bank. If it is pursuing the lending policy described above logically, it is in a position to

make ongoing adjustments to the debt rate that are appropriate to the situation. If the company’s

revenues fall so far that its market value becomes identical to its liquidation value, then the

result is logically, on the basis of the expositions in this section, a maximum debt rate close to

zero in relation to liquidation value. Thus the bank has no problem.

The situation is quite different if a sharp drop in earnings occurs very suddenly. The company

is then no longer in a position to make the repayments that would be required to reduce the

debt rate to the necessary level concerned. Thus there arises a case for the bank to wind up.

9.7.5 Consequences for Loan Supervision

Derived from the expositions in this section so far, the following tasks arise for the bank in the

context of loan supervision:

r

supervision of the free cash ﬂows of the company seeking loans

r

derived therefrom, supervision of the company’s market value and of its volatility and of its

debt rate

Combination of Loans 115

r

regular calculation of the maximum debt rate appropriate to the maximum return on equity

(see Figure 9.5)

r

checking whether repayments have to be demanded in order to get back to the safety interval

away from the maximum debt rate, as deﬁned as a matter of business policy

r

laying down fresh risk-adjusted loan conditions

r

handing over care of the company seeking the loan to the bank’s winding up department,

to the extent that the company may no longer be able to make the necessary repayments, if

required.

It must be any bank’s objective that winding-up cases only arise, if the sharp drop in revenues

that befalls the company seeking the loan comes so rapidly that the repayment requirements

arising necessarily from that are no longer acceptable to the company. In all other cases the

necessary repayments should be called in consistently, as explained, in order to protect the

bank from damage.

9.8 RESULTS AND CONCLUSIONS

The marginal interest rate of a loan plays a central part in the calculation of the ﬁnancial

structures that are most favourable in terms of interest costs, and in the determination of the

maximum debt rate:

i

m

= i (L) + L ·

∂

∂L

i (L) (9.4)

Several loans to the same company are structured most favourably in terms of interest costs

if all marginal interest rates are identical:

i

mk

= i

mn

for all k with 1 ≤ k ≤ n −1 (9.20)

The maximum achievable return on equity for a company corresponds to the marginal

interest rate described above, where the debt structure is appropriate.

e = i

m

(9.32)

Companies with a debt rate that is higher than that of the maximum return on assets are in

very serious danger of going bankrupt. This therefore is also the maximum debt rate at which

the bank should still be providing ﬁnance. It is, however, recommended not to go so far in

the provision of outside ﬁnance that this situation is reached, in order to be able to prevent

bankruptcy in the case of sharp falls in earnings occurring.

As long as a company has no more outside capital than matches the debt rate for maximum

return on equity, the interest is still acceptable. The maximum acceptable repayments corre-

spond to the return on equity after taxation. The maximum debt rate, particularly in the case

of the ﬁnancing of far-reaching growth, is thus also bound up with the question of whether

the return on equity repayments permits the possibility of the debt rate returning ‘sufﬁciently

rapidly’ to a ‘bankruptcy-resistant’ debt rate (see subsection 9.7.4).

In order to avoid cases of bankruptcy where there are sharp drops in earnings as outlined

(Section 9.7), the bank must consistently call in the necessary repayments. If it does not do so,

then bankruptcy is as good as inevitable.

116 Risk-adjusted Lending Conditions

9.9 EXAMPLE

The example in subsection 7.12.1 may again serve as the basis. It is intended that the loan of

1000 over three years should, however, nowbe guaranteed, not as in the example in Section 8.6

by a mortgage on a building, but by two mortgages on two different properties, on the most

favourable conditions in terms of interest costs. The two properties have the characteristics

shown in Table 9.1.

This results in the following values for the interest curve, whereby only that part of the table

that is interesting for the purposes of the example is shown in Table 9.2. The credit risks ρ

∗

result from the application of equations (7.37)–(7.38). The risks ρ

∗

B ∩C

are again the product

of ρ

∗

times 0.1113 according to Table 7.6 as in the example in Section 8.6. The rate of interest

i is calculated according to equation (4.49), whereby the risk-free rate of interest according

to Table 7.6 again amounts to 4.5%. The marginal interest rate is calculated according to

equation (9.5). The total interest results from the sum of the two mortgage amounts per line,

times the loan interest rate concerned. (Please note: if the ﬁgures in the following table are

checked, rounding differences may appear. The ﬁgures were originally calculated using an

Excel worksheet to an accuracy of several more decimal points.)

The total incidence of interest is at its smallest (48.42), if House A is mortgaged at 460

and House B at 540. At these ﬁgures the difference between the two marginal interest

rates is also, as one would expect, at its lowest. This now leads to the ﬁnancing shown in

Table 9.3.

The difference between the two rounded up rates of mortgage interest amounts to

5

8

%. This

shows clearly that not just the rates of interest, but the marginal interest rates, must agree with

each other, in order to achieve optimum ﬁnancing.

Table 9.1 Starting out position

Market value Volatility of market value

House A 800 50%

House B 600 10%

Table 9.2 Marginal interest rates

House A House B

mortgage ρ

∗

(%) ρ

∗

B ∩C

(%) i (%) i

m

(%) mortgage ρ

∗

(%) ρ

∗

B ∩C

(%) i (%) i

m

(%) Total interest

430 3.98 0.44 4.96 6.91 570 3.11 0.35 4.86 13.65 49.07

440 4.38 0.49 5.01 7.14 560 2.09 0.23 4.74 10.32 48.62

450 4.80 0.53 5.06 7.38 550 1.41 0.16 4.66 8.37 48.43

460 5.25 0.58 5.11 7.63 540 0.94 0.11 4.61 7.11 48.42

470 5.73 0.64 5.17 7.90 530 0.62 0.07 4.57 6.26 48.54

480 6.24 0.69 5.23 8.19 520 0.40 0.04 4.55 5.67 48.75

490 6.78 0.75 5.29 8.49 510 0.25 0.03 4.53 5.27 49.04

Combination of Loans 117

Table 9.3 Final results

House A House B

Mortgage 460 540

Risk ρ

∗

0.58% 0.11%

Rating BB A

Rating risk (Table 2.1) 0.757% 0.171%

Risk-free rate of interest 4.5% 4.5%

Mortgage interest rate 5.297% 4.697%

Mortgage interest rate rounded up 5 5/16% 4 11/16%

That House B in this example —in the case of ﬁnance that is at its most favourable in terms

of interest costs —is mortgaged at a much higher rate than House A, although its market value

is signiﬁcantly lower, is connected with the much lower volatility of the market value of House

Bin comparison with House A. The ﬁgures for this example were deliberately chosen to obtain

results that would be as illustrative as possible.

Part IV

Implementation in Practice Implementation in Practice

Procedure —according to the model —for assessing the risk

in lending to a company

Applications

Final considerations

10

Procedure —According to the

Model —For Assessing the Risk in

Lending to a Company

In the normal course of events representatives of the lending bank discuss their business

with representatives of the loan receiving company at least once per annum, with the aid

of annual and/or intermediate accounts. We will demonstrate in this chapter how to proceed

in this, in order to be able to ﬁnd out about the necessary model parameters as reliably as

possible.

10.1 OVERALL VIEW OF THE PROCEDURE

The following values have to be ascertained using equations (7.37) and (7.38): debt rate,

volatility of assets and term of loan.

The debt rate is worked out as the quotient of the outside capital divided by the value of the

assets. The value of the outside capital is taken from the balance sheet. The value of the assets

is determined, according to subsection 7.4.1, on the basis of the discounted future free cash

ﬂows. The free cash ﬂows achieved in the past provide a clue for these, which must in turn

support a forecast of the future asset values. The discount rate results from the CAPM. As was

explained further in subsection 7.4.1, the value of the assets may never be smaller than their

liquidation value. A check must therefore be made as to which of the two values is the greater

(discounted free cash ﬂows or liquidation value).

The volatility of the assets results from analysis of a range of annual accounts and budgets.

It follows from this that there must be a track record in terms of annual accounts for there to

be any successful loan analysis. Budgets should be brought into the calculations too, in order

to take account of inﬂuences that may come to bear on volatility in the future.

The following documentation must therefore be brought together for the loan assessment:

r

Acomplete set of the company’s past accounts, it being unimportant whether they are annual,

six-monthly or quarterly accounts, as long as they have been drawn up according to the same

principles. An important question here is how far back into the past to delve. One answer

to this question is —‘as far as possible, in order to obtain the broadest possible statistical

basis’. Another answer is: ‘only so far as the company’s past operational situation lines up

with its current operational situation, in order to avoid bringing into the assessment any

factors that are now no longer relevant’. So the answer to the question of how many sets of

accounts from the past should be included in the assessment is far from trivial, and does in

fact have to be answered afresh on each occasion.

122 Risk-adjusted Lending Conditions

r

Reliable budgets for the future. Here more value should be ascribed to reliability than to any

far sight into the future. Clear ideas ought, however, to exist at least for the current and for the

following ﬁnancial year. If not, one is entitled to raise the question of whether the company’s

management has sufﬁcient specialist capabilities for the company to be credit-worthy at all.

r

The necessary ﬁgures, according to CAPM, for determining the discount rate: return on the

market, yield of a risk-free investment in government bonds, and the extent to which the

non-mortgaged assets are at risk in the marketplace.

r

The remaining terms and notice periods of loans that have already been granted.

In the following sections we will on the one hand explain how this documentation should

be analysed. On the other hand we will go, in each case, into what values are critical for the

analysis and which must therefore be probed appropriately in discussion with the company’s

representatives.

10.2 ANALYSIS OF EARNINGS STATEMENTS

As mentioned in the preceding section, the determination of free cash ﬂow is involved in the

analysis of earnings statements. According to Brealey and Myers [BRMY96, S. 71/72] the

following applies:

Free cash ﬂow = revenues − costs − investments

The task is thus set, of determining sales, operating costs and investments. Here it must be

ensured that the operating costs do not contain any depreciation or interest on debt, as it is the

value of the company’s assets that have not been mortgaged that has, according to the model,

to be determined.

In portraying sales in earnings statements it is often necessary to establish that distinctions

are made between operational and non-operational, and between ordinary and extraordinary

revenues. Such distinctions are of no importance at this point. Once the company’s management

has decided that the enterprise should embark on certain activities, these then become an

element in the company’s activity and therefore formpart of its risk proﬁle. So all the company’s

revenues must be added together when it comes to determining revenue.

The same applies for costs as explained above in respect of revenues. All costs with the

exception of depreciation, provisions and interest on outside capital must be taken into account.

Other ﬁgures in earnings statements are irrelevant in connection with analysing loan risks.

In particular depreciation and provisions are of no interest in this context, as they do not feature

in the calculation of free cash ﬂow.

When discussing accounts it should be ensured that the complete sales and operating costs

for all existing accounts and budgets can be found out as precisely as possible. Net positions

should be subdivided again into their individual elements, as the result is otherwise distorted.

10.3 ANALYSIS OF CASH FLOW STATEMENTS

The amount of investments is still missing for determining free cash ﬂow. This is worked out

fromthe cash ﬂowstatement, and here all investments must be taken into consideration in turn.

Assessing the Risk in Lending 123

And here too, as already explained in the preceding section, no possible distinctions should be

preferred according to the various categories.

When discussing accounts it should therefore be ensured that the company’s previous and

planned investment activity can be found about comprehensively. All the ﬁgures necessary for

calculating the free cash ﬂows are thus determined.

10.4 ANALYSIS OF BALANCE SHEETS

Under the model, the value of a company’s assets are worked out either from the value of the

discounted free cash ﬂows or from the liquidation value of the assets, whichever value is the

larger. It follows from this that in respect of the assets only their possible liquidation value

in the event of bankruptcy should be determined. Or, put another way —neither the current

operational value of the assets, nor the company’s depreciation and provision policy, are of

any signiﬁcance for the analysis of credit risk.

On the liabilities side of the balance sheet it is only the current and planned debts budgeted

vis-` a-vis third parties that are of interest, in order to be able to determine the current and future

debt rate. Here any possible categories of loans must be taken into account. In order to evaluate

privileged claims correctly, subordinated debts such as equity must therefore be considered. It

follows from this that the calculation of the debt rate must be undertaken separately for each

loan, if need be according to which category it falls into. Here only debts in the category having

precedence or in the same category have to be taken into account.

When discussing accounts it should therefore be ensured that the liquidation value of the

assets can be found out about as precisely as possible. If the debts are proved, it may be

assumed that these are correctly stated. If they are not, then it could be a case of falsiﬁcation

of documents.

10.5 DETERMINATION OF THE DISCOUNT RATE

In order to determine the respective value of assets, the discount rate applicable at that moment

must be determined for each set of accounts, for the discounting of the free cash ﬂows. Howthe

CAPM may be used for determining the discount rate was explained in summary in subsection

7.4.1. The determination of the risk measurement β plays a decisive role here. The relevant β

can be calculated as follows in the case of companies quoted on the stock exchanges [BRMY96,

S. 215]:

β

V

=

D

V

· β

D

+

E

V

· β

E

(10.1)

β

V

= risk measurement of the value of the company

β

D

= risk measurement of the debts

β

E

= risk measurement of the equity

It may be assumed here that the values for β

E

and β

D

are available from suppliers of

information such as Bloomberg or Reuters. Should a company have only quoted shares and

124 Risk-adjusted Lending Conditions

not bond loans as well, then the value for β

D

should be supported by values of companies in

the same or related sectors.

The determination of β is more difﬁcult in the case of companies that are not quoted. One

solution here is to use the calculation of the β-values of quoted companies in the same or related

sectors, as described above. Such comparisons are admittedly not simple in Switzerland, as

only a few companies are quoted compared with larger countries. Foreign values can possibly

be used by way of comparison too. This question too may only be answered, at the end of

the day, by empirical testing. The higher β-values should be used in comparisons, under the

principle of conservatism. Otherwise we refer, at this juncture, to appropriate reading (for

example [BRMY96, Chapter 9: Capital Budgeting and Risk]).

10.6 DETERMINATION OF RELEVANT LOAN TERMS

When determining the relevant term for the loan using the model, the question comes up of

how long will it be until the bank granting the loan gets its money back.

The relevant loan terms result primarily from the loan agreements. In the case of ﬁxed-

rate loans, it is a question of the residual term until maturity. In the case of all other loans,

the notice period is deﬁnitive. Here it must, however, be ensured that critical situations are

normally recognised only following a periodic discussion of accounts. Moreover, the duration

of the proceedings in any possible bankruptcy must be taken into account. So the following

terms ensue, which must be inserted into the model:

r

In the case of ﬁxed-rate loans, the term relevant to the credit risk corresponds to the residual

term plus the possible duration of any bankruptcy proceedings.

r

In the case of loans on which notice can be given and of no ﬁxed duration, the term relevant

to the credit risk corresponds to the sum of the time until the next discussion of accounts,

plus the notice period, plus the possible duration of any bankruptcy proceedings.

It follows from this that the volatility that is dependent on the term of the loan, and thus the

credit risk, may be lowered to the extent that the terms of ﬁxed-interest loans, the periods of

notice, or the intervals between discussions on accounts are kept short.

10.7 DETERMINATION OF VOLATILITY

The volatility of the company’s value may be calculated, according to subsection 7.4.2, on the

basis of the values calculated —per set of accounts and per budget —for the free cash ﬂows,

for the discount rates and for the liquidation values.

10.8 DETERMINATION OF CREDIT RISK

As soon as all values have been calculated as described in the preceding sections, the shortfall

risk of each of the company’s loans can be determined as described in Section 7.3.

The method will be illustrated in Chapter 11, with the aid of three examples.

Assessing the Risk in Lending 125

4.00%

3.00%

2.00%

1.00%

0.00%

0 5

Mortgage 75%/Volatility 10%

10 15

Probability of bankruptcy in the

respective year of the term of the loan

Figure 10.1

10.9 PRUDENCE IN THE CASE OF NEW LOANS/BORROWERS

It has beenpossible toobserve, empirically, that default onloans occurs muchmore frequentlyat

the beginning of the term than after a certain duration (see, for example, [FRRM97, S. 112]).

That is to say loans which can already point to some duration are much safer for the bank

granting them than newly approved loans.

This fact can be understood from the model described here. (The values for the mortgage

and volatility were selected for Figure 10.1 in such a way that the resulting curve is about the

same as in [FRRM97, S. 112].)

It is therefore advisable to have particular caution prevail in the case of new business

transactions, such that ﬁgures based on pessimistic expectations are inserted into the model.

It is furthermore recommended that new loans are checked more frequently at the beginning

of their terms.

The recognised rating agencies behave in just this way when it comes to assessment of

asset-backed and mortgage-backed securities. It is an advantage for a bank to put together for

securitisation portfolios of loans in which the loans already have higher seasoning. In these

cases the rating agencies demand smaller loan reinforcements for the achievement of a top

rating than for portfolios with newly granted loans.

10.10 POSSIBLE CAUSES OF CONFLICT BETWEEN BANK

AND BORROWER WHEN THE MODEL IS APPLIED

The application of the model described demands a fair amount of specialised knowledge and

experience, not only from the bank personnel concerned. Demands are made on borrowers,

too, in that they are tending to have to furnish more, and more detailed information about

126 Risk-adjusted Lending Conditions

their companies. This requires appropriate knowledge, on top of willingness to share. Such

operational knowledge may, however, not be presupposed to exist in all small and medium-

sized enterprises. It follows from this that auditors commissioned to produce annual accounts

have to receive broadened mandates, which entails matching extra costs.

As was explained in Section 9.6 and illustrated in Figure 9.5, the response on the bank’s

part to any company having a high debt rate must be the application of appropriately higher

loan interest rates. If the rules described in Chapter 9 are kept to, these interest rates are

also acceptable to the borrower and still make operational sense because it may increase its

return on equity without running irresponsible risks. Such high-risk premiums on loan interest

rates —as calculated by the model in some circumstances —have, however, been unusual in

Swiss banking practice hitherto. Readiness to pay such interest rates if appropriate still has

to be stimulated. This will be all the more difﬁcult, and take all the longer, while other banks

(still) do not apply loan conditions that are consistent with risk.

Particularly for the two reasons outlined above, application of the model may lead to conﬂict

between bank and borrowers:

r

The scope of the information required will be seen as an unreasonable demand.

r

The additional costs for auditors will be viewed as pointless expenditure.

r

The high loan interest rates will seem to threaten commercial viability, even if that is not the

case.

These causes of conﬂict should be countered by careful training of the bank ofﬁcials con-

cerned and where applicable the borrowers, by means of detailed explanations. It does of course

go without saying that this is far from simple. The application of risk-adjusted loan conditions,

combined with the careful examination of credit-worthiness required for that, is, however,

imperative for the continued existence of any bank. The commercial difﬁculties experienced

by some banks at the beginning of the 1990s, not least owing to far-reaching ﬁnancing of

property and the excessively low mortgage interest rates that were associated with it, and to

insufﬁciently careful checks on credit-worthiness, have clearly proved this [BRUN94, S.137

and thereafter].

11

Applications

In this chapter, ﬁrst of all, we will investigate three practical cases and compare the results

with the conventional assessment made by the lending bank concerned. All ﬁgures we have

given are not the actual sums in Swiss francs that were concerned, thus making it impossible

to draw conclusions about the identity of the bank customers involved and maintaining bank

conﬁdentiality. The proportions that the ﬁgures in any example bear to each other do, however,

match, in order to make for accurate analysis. Only in Section 11.3 may checking come up

with some minor discrepancies, as the change in the starting values led to smaller changes in

the proportions, while the results were calculated using the original ﬁgures.

11.1 SPECIALIST CLOTHING BUSINESS: TURN-AROUND

SITUATION

This example is about an established business specialising in ladies’ and gentlemen’s clothing,

with shops throughout Switzerland, in a turn-around situation. Annual accounts for the last

ﬁve years, and budgets for the current year and for the next two years ahead, were available

for analysis. Those in the bank responsible for the customer considered the company’s budgets

very reliable, on the strength of their previous experience, so these were used for the analysis.

First the EBIT for the individual years is calculated and then capitalised using the long-term

market return of 7%, whereby it is assumed implicitly that a value of β = 1 is, according to the

CAPM, applicable in this case. Then the company’s liquidation value for the individual years

is determined from its assets, using normal banking correction factors. The higher of earning

capacity value and liquidation value is taken as the market value for further calculations.

The applicable debt rate is calculated with reference to outside capital owed to third parties.

An approximate ﬁgure of 20% of annual wages and salary costs is inserted for privileged

salary claims in the event of bankruptcy, so that the corrected loan risks are calculated (see

Section 7.9): the ﬁgures below are obtained, if one inserts 4% for the standard rate of interest.

Volatility is worked out using equation (7.45) on the basis of the company’s market value.

Loans are in each case granted for one year. The Excel worksheet in Appendix 2 is used for

the calculations, and the results are summarised in Table 11.1.

In recent years this customer was classiﬁed by the bank as an increased credit risk, which is

fully understandable on the basis of the above analysis. A new management team, assessed as

competent, took over responsibility for the company at the beginning of the 1990s, and in 1995

an external ‘turn-around’ adviser was brought in as well. The loan was therefore continued, in

the hope of better times.

The budgets for 1999 to 2001 are encouraging, and on the strength of their past reliability, the

turn-around expected for 1999 seems actually to be in sight. So those responsible for the client

128 Risk-adjusted Lending Conditions

Table 11.1 Specialist clothing business: model information and results

Five annual accounts Three budgets

Year 94 95 96 97 98 99 00 01

EBIT 70 84 66 76 74 100 106 122

Earning rate 1000 1200 943 1086 1057 1429 1514 1743

Breakup value 1060 960 920 800 820 820 820 800

Market value 1060 1200 943 1086 1057 1429 1514 1743

Debts 1076 1048 1000 972 908

Debt rate (in %) 99.1 99.1 70.0 64.2 52.1

Privileged salaries 56 58 60 60 60

Volatility of the market value: calculated: 17.7%

rounded: 20.0%

∗

Annual credit shortfall risk (in %) 24.5 24.3 0.37 0.12 0.004

Corrected annual credit shortfall risk (in %) 49.0 49.6 0.66 0.22 0.009

Rating according to Table 2.1 DD DD BB BBB AAA

∗

Volatility was rounded up to 20% to take uncertainties into account.

NB: The ﬁgures for 1994 to 1996 were not calculated out, being of no further interest by spring 1999.

have applied to the loan assessment ofﬁce for the client to be assigned to a higher category of

credit-worthiness within the bank. As this application is only supported by budgets, it seems

premature. Before anything is undertaken on this, the positive prospects should at least be con-

ﬁrmedbymeans of a set of intermediate accounts as at 30.6.99. Suchintermediate accounts have

been being obtained for a long time, anyway, because of the strained credit-worthiness situation.

This example allows us to show, impressively, that loan risk assessment must be underpinned

by parameters set in the future because —on the basis of the current situation —the loans to

this company would have had to be terminated without notice. The positive prospects for the

future do, however, allow their continuation to appear appropriate. When it comes to setting

the interest rate here, this may neither be based on the current situation (too pessimistic) nor

on the budget projection (currently as yet too optimistic). This combination of circumstances

calls, rather, for the interest rate to be set on the basis of what the company can accept to pay.

An EBIT for 1999 of 100 is budgeted for distribution. Equity as per the balance sheet of

31.12.98 comes to 82. The earnings reported for 1999 should therefore be split into 100 parts,

of which 5 remain as earnings and 95 are set aside for debt servicing. The bank-related debt

amounts to 938, the remaining debt being non-interest-bearing positions such as creditors. An

average rate of interest on debt of 10% is realistic for the bank in the context of the budget

being attained. But in order not to jeopardise the turn-around by imposing excessively high

interest costs, at least a part of the loans might be converted into a proﬁt participating loan

with returns thereon being dependent on success.

The conventional loan decision and the loan decision emerging from the use of the model

are qualitatively identical in this example. The advantage of the model consists, however, in the

fact that the risks involved are quantiﬁable and appropriate loan supervision measures may be

laid down. The uncertainties regarding the quality of budgeting, the development of the market

and of the sector, and the capabilities of the management team were taken into account with

Applications 129

an increase in volatility. Here it is once again shown that establishing the ﬁgure for volatility

is of decisive importance. The banker’s experience plays an important part in this.

11.2 COMPANY TRADING IN MACHINE TOOLS: PROVISION

FOR SUCCESSOR COMPANY

The two major shareholders in a company trading in machine tools each own 48.5% of its

share capital. They form the management team, together with a third person who owns the

remaining 3% of the shares. The older of the two major shareholders would like to withdraw

from the business, and offers the younger his shares at a price of 240. The latter intends to

form a holding company to acquire the shares. He applies to the bank seeking a loan for the

full purchase price, whereby he would pledge the entire shareholding of 97% in his ownership

in favour of the loan. He intends, on the basis of a freshly drawn up business plan, to repay the

loan over ﬁve years, with payments of 48 per annum.

Only the 97% of the future dividends of the trading company will be available to service the

debt. Disregarding minor management costs, this dividend income corresponds to the com-

pany’s EBIT. The loan application has therefore to be examined on this basis. The three most

recent sets of annual accounts and ﬁve forward budgets of the trading company are available

for this examination. EBIT will again be capitalised at 7%in order to determine the company’s

market value. The liquidation value of the company matches the current value of the shares.

Analysis of the ﬁgures made available has resulted in the company’s market value in all years

having been higher than its liquidation values.

The relevant ﬁgures are shown in Table 11.2 (calculations with the help of the worksheet in

Appendix 2).

An average loan risk of 5.1% results, based on the 1997 annual accounts. With a standard

rate of interest of 4%, this makes a loan interest of about 10%. This leads in 1998 to a debt

Table 11.2 Trading company: model information and results

3 Annual accounts Five budgets

Year 95 96 97 98 99 00 01 02

∗

Holding-EBIT 15 13 52 93 97 98 101 98

∗

Market value 214 186 743 1329 1386 1400 1443 1400

Market value volatility 56.8%

rounded 60.0%

loan 240 240 192 144 96 48

∗∗

Duration loan 3.0 3.0 2.5 2.0 1.5 1.0

Credit shortfall risk (t) (in %) 14.7 5.1 1.64 0.28 0.007 0.0

Average annual credit shortfall risk (in %) 5.1 1.7 0.66 0.14 0.005 0.0

Rating according to Table 2.1 CC CCC BB A AAA AAA

∗

Corresponding to 97% of the dividends of the trading company.

∗∗

Disregarding interest.

NB: The ﬁgures for 1995 and 1996 were not calculated out, being of no further interest by spring 1998.

130 Risk-adjusted Lending Conditions

service of 24 (interest) + 48 (repayment) = 72. This is acceptable under the 1998 budget, but

not in relation to the 1997 accounts. Any reduction in the repayment would, conversely, lead to

an extension in the duration and thus increase the loan risk and interest rate. That is therefore

no solution. The budgets do, moreover, seem really rather optimistic, viewed in relation to

the past. The loan application should therefore be declined on the grounds of questionable

acceptability.

It is apparent from this example that budgets that are very optimistic compared with past

performances do lead to high volatility in a company’s market value. In this case, considerable

loanriskwouldhave ensuedfromminor mortgagingof just about one-third(240/743 = 0.329).

Related to any operating company, this means that any fast growth must be ﬁnanced above

all by its own cash ﬂow, because of the high volatility associated with it: ﬁnance using bank

loans becomes rapidly associated with high risks that lead, in turn, to matching high (too high!)

interest costs.

For this loan application the bank had also established by conventional methods that it should

refrain from granting the loan. The business was not done. The questionable acceptability was

also decisive. The budgets were assessed as being too optimistic when it was taken into account

that an experienced member of the management team was leaving the company.

The qualitative loan decision reached both by conventional means and via the model was

identical in this example too. The model does, however, deliver detailed reasoning for having

to turn down the loan application.

11.3 SHIP MORTGAGES: RISK LIMITATION

This example is all about ﬁguring out where the risk in the loan transaction lies.

11.3.1 Starting Position

Amajor shipowner has had a group of six bulk tankers ﬁnanced by a consortiumof eight banks.

Ship mortgages have been set up for this purpose. The ships were built in 1976/77. As a result

of costly special surveys the value of the ships, according to estimates by two internationally

respected ﬁrms of shipbrokers, was put at a total of 1000 at 29.9.98. The scrap metal value of

the ships at the same point in time amounted to 212. The loans were 378 at 30.6.98.

The shipowner applied in the middle of 1998 to increase the loans to 450, in order on the one

hand to ﬁnance reﬁtting costs and on the other to have liquid funds available for the purchase

of new ships. The banks granted the loan application on the strength of the low proportion of

the ships’ value mortgaged and of the fact that it had been possible to extend their working

lives through to the middle of 2003, thanks to the reﬁts. Cash ﬂow analysis of future revenues

worked out that ships’ incomes of about 0.11 per day at sea would sufﬁce to cover the costs of

operating and maintaining them, and the loan costs.

11.3.2 The Banks’ Loan Decision

The banks consented to the increase in the loans and agreed with the shipowner the repayment

plan in Table 11.3. Due note was taken that the loan outstanding of 171 as at 30.0.03 would

Applications 131

Table 11.3 Ship mortgages: model information

From end Debt rate High Low

of quarter Amortisation Loan Depreciation Ship value (in %) volatility (in %) volatility (in %)

III 98 450 1000 45.0 43.0 27.0

IV 98 13.1 436.9 41.5 958.5 45.6 41.8 26.2

I 99 15.2 421.7 41.5 917.0 46.0 40.6 25.4

II 99 19.8 401.9 41.5 875.5 45.9 39.4 24.6

III 99 13.6 388.6 41.5 834.0 46.6 38.2 23.8

IV 99 13.6 374.7 41.5 792.5 47.3 36.9 23.1

I 00 13.6 361.1 41.5 751.0 48.1 35.7 22.3

II 00 18.0 343.1 41.5 709.5 48.4 34.5 21.5

III 00 13.6 329.5 41.5 668.0 49.3 33.3 20.7

IV 00 13.6 315.9 41.5 626.5 50.4 32.1 19.9

I 01 13.6 302.3 41.5 585.0 51.7 60.9 19.1

II 01 18.0 284.3 41.5 543.5 52.3 29.7 18.3

III 01 13.6 270.7 41.5 502.0 53.9 28.5 17.5

IV 01 13.6 257.1 41.5 460.5 55.8 27.3 16.7

I 02 13.6 243.5 41.5 419.0 58.1 26.1 15.9

II 02 18.1 225.4 41.5 377.5 59.7 24.8 15.2

III 02 13.6 211.8 41.5 336.0 63.0 23.6 14.4

IV 02 13.6 198.2 41.5 294.5 67.3 22.4 13.6

I 03 13.6 184.6 41.5 253.0 73.0 21.2 12.8

II 03 13.5 171 41.0 212 80.7 20.0 12.0

certainly have been covered by the scrap metal value of the ships. The positive decision was

above all underpinned by the following considerations:

r

At 45%, the proportion of value mortgaged was small at the time of the increase in the loans.

r

The loan repayment at the end of the ships’ lives —the ‘balloon’ payment —is well covered

by the scrap metal value of the ships that was expected at that time.

r

According to the bank’s calculations the break-even freight rate amounts to about 0.09 per

day, based on details supplied by the shipowner and on comparison with similar ships —the

current freight rate of 0.11 per day was thus quite well sufﬁcient.

r

The ship owner’s management team was judged to be very capable and reliable.

11.3.3 Assessment of the Loan Risk by the Banks

The bank that made the documentation available had assessed the loans with a loan risk of

80 basic points according to its internal rating system. This classiﬁcation was made on the

strength of comparisons with loans in similar positions.

11.3.4 Determination of Loan Risk According to the Model

Additional factors for determining loan risks according to the model must include the volatility

of ship prices and of scrap metal prices, realisation costs in the event of bankruptcy, time scales

and ship depreciation:

132 Risk-adjusted Lending Conditions

Volatility

Details from the spring 98 issue of the bi-annual Clarkson Shipping Review and Outlook —

from the Clarkson ﬁrm of shipbrokers —were used for determining the volatility. Both the

time series from 1979 to 1998 for the tankers in question (Page 16, column 280k 1975) and the

relevant scrap metal prices (Page 153) are to be found there. The following volatility ﬁgures

for each of the years concerned may be calculated by using this information:

Ship prices: 1979–98: 43%

1993–98: 27%

Scrap-metal prices: 1979–99: 20%

1993–98: 12%

Behind the use of two different time periods is the consideration that the 1979–98 sequence of

numbers allows calculation of the worst case, and the ﬁve-year 1993–98 period is most likely

to be representative for the last part of the termof the loans. As it was possible to depreciate the

ships using the straight-line method (see below), it may be assumed for the sake of simplicity

that volatility of the ship prices moves in line with volatility of the scrap-metal prices (see

Table 11.3).

Realisation Costs in the Event of Bankruptcy

These amount to about 4.5 per ship, according to the bank’s information. Costs of about 27

have therefore to be allowed for six ships.

Time-scale Considerations

The loan agreement contains a clause covering the event of default. Thus, if the shipowner is

at any point not in a position to service the debt on time, then the consortium of banks has the

right to make the loan repayable immediately and to realise the value of the ships. According

to the bank’s information, such ships can be realised within three to six months, be it privately

or by means of auction. The loan may be utilised on a roll-over basis in the form of advances

in Euro of, in each case, terms of from one to 12 months. The longer-term advances have

intermediate interest deadlines, in each case after three months. So every three months the

banks may establish whether the shipowner is still in a position to service its debts. On top

of the time needed for realisation of the ships, a reaction time of from six to nine months is

worked out for the banks. The following risks are examined in terms of their time-scale aspects

on the basis of this situation:

r

High volatility: 6, 9, 12 and 15 months from the most recent proper payment service the

debt.

r

Lowvolatility: 6, 9, 12 and 15 months fromthe most recent proper payment service the debt.

Depreciation of the Ships

According to the bank’s information, the ships may be depreciated using the straight-line

method from their current value of 1000 as at 29.9.98 to their scrap value of 212 as at 30.6.03.

Applications 133

Table 11.4 Ship mortgages: results in %

Low volatility High volatility

Whole term 0.5070% 2.8073%

Term Term

From end

of quarter 6 months 9 months 12 months 15 months 6 months 9 months 12 months 15 months

III 98 0.0004 0.0077 0.0318 0.0749 0.2025 0.6320 1.1287 1.6099

IV 98 0.0003 0.0063 0.0271 0.0653 0.1795 0.5751 1.0398 1.4939

I 99 0.0002 0.0048 0.0219 0.0545 0.1530 0.5091 0.9373 1.3610

II 99 0.0001 0.0030 0.0151 0.0401 0.1156 0.4142 0.7904 1.1724

III 99 0.0001 0.0024 0.0127 0.0346 0.1022 0.3762 0.7273 1.0868

IV 99 0.0001 0.0021 0.0113 0.0314 0.0886 0.3369 0.6613 0.9968

I 00 0.0001 0.0017 0.0097 0.0276 0.0797 0.3101 0.6148 0.9320

II 00 0.0000 0.0011 0.0069 0.0207 0.0617 0.2569 0.5265 0.8135

III 00 0.0000 0.0009 0.0060 0.0185 0.0568 0.2404 0.4961 0.7692

IV 00 0.0000 0.0008 0.0054 0.0169 0.0537 0.2294 0.4746 0.4368

I 01 0.0000 0.0008 0.0051 0.0161 0.0527 0.2245 0.4634 0.7179

II 01 0.0000 0.0005 0.0038 0.0124 0.0427 0.1927 0.4085 0.6423

III 01 0.0000 0.0005 0.0039 0.0129 0.0460 0.2013 0.4200 0.6357

IV 01 0.0000 0.0007 0.0046 0.0146 0.0537 0.2230 0.4525 0.6921

I 02 0.0000 0.0010 0.0062 0.0184 0.0699 0.2669 0.5158 0.7719

II 02 0.0000 0.0011 0.0069 0.0202 0.0716 0.2705 0.5216 0.7721

III 02 0.0001 0.0028 0.0137 0.0353 0.1266 0.4027 0.7119 0.9997

IV 02 0.0007 0.0102 0.0373 0.0802 0.2839 0.7136 1.1211 1.4655

I 03 0.0095 0.0607 0.1487 0.2512 0.8509 1.5808 2.1343 2.5463

The model information needed according to the above expositions may be taken from

Table 11.3. The results are summarised in Table 11.4. It was assumed, in the case of the

calculations of the quarterly loan risks, that the situation as it was then is presented as in

Table 11.3, and that the shipowner had up until then met its commitments fully on each

occasion. The calculations were made with the aid of the worksheet in Appendix 2.

As may be seen in Table 11.4, the loan risk is heavily dependent on the bank’s reaction

time. The ﬁgures in the case of 15 months are signiﬁcantly higher than for six months, under

the assumption both of the lower and of the higher rates of volatility. It becomes apparent that

assessment of the volatility of the value of the collateral and the bank’s reaction time acquire

decisive signiﬁcance in the case of this loan.

It is more difﬁcult to frame the risk calculation over the full term. The assumption was ﬁrst

of all made that servicing the debt had until then taken place in full at the beginning of any

calendar quarter. It had therefore been worked out at ﬁrst, for calculating the loan risk, how big

it would have been on the assumption that the remaining debt was still outstanding at that point

in time. Then the same loan risk was calculated as at the end of the quarter, and the loan risk

was determined for that quarter from the two ﬁgures taken together. The weighted average, in

terms of amount, of these risks per quarter, converted to one year, then give the loan risk per

annum over the entire term (Table 11.4).

134 Risk-adjusted Lending Conditions

A situation assessment has to be carried out ﬁrst of all, because of the high inﬂuence of the

bank’s reaction time on the loan risk in the event of default. The amount of the loan, the current

value of the ships and the volatility of the development in value may not be inﬂuenced. On the

other hand, realisation of the ships may or may not be forced in terms of time, depending on

the risk situation. So it has to be assessed whether it is more advisable for the bank to realise a

lower price in the short term, or whether it should attempt to achieve, over a somewhat longer

time-scale, a price that is higher but still uncertain.

11.3.5 Comparison of Assessment between the Bank and the Model

There is a credit shortfall risk of about 51 basic points on the assumption of the lower rate of

volatility, considered over the full term of the loan. But as every three months it can be viewed

again whether the shipowner is or is not still meeting its commitments, this is a worst-case

scenario. Seen from the point in time at which the loan is discussed there results, for each

individual quarter for three or six months (i.e. including the time expected for realisation) a

normal credit shortfall risk of less than one basic point. Only for the last two quarters are the

ﬁgures correspondingly higher, because of the higher proportion of mortgage existing at that

point, but they are still very small. It follows that the bank assessment is too pessimistic on the

assumption of the lower rate of volatility.

For reasons of conservatism it may be argued that a credit rating must be aligned to a longer

than normal realisation period. As may be gathered fromTable 11.4, however, loan risks that are

still less than ﬁve basic points result from reaction times of 12 and 15 months (corresponding

to realisation periods of 9 and 12 months), except in the last two quarters.

There is a credit shortfall risk of about 2.8%on the assumption of the higher rate of volatility,

considered over the whole term of the loan. It is only here that the decisive inﬂuence of the

volatility ﬁgures used becomes apparent. But this too is again a worst-case scenario. From the

point of view of the individual quarters, and with the exception of the last quarter, ﬁgures of

from 4 to 72 basic points emerge for the two shorter reaction times. In the ﬁnal quarter the

ﬁgures are again somewhat higher. It follows from this that the bank assessment lines up more

or less with the ﬁgures for the higher rate of volatility taken from the model. Only in the case

of reaction times of one year or more would the risk have to be estimated as being higher.

It follows from the above reﬂections that the bank assessed this loan very cautiously. When

considered in this sort of way, the bank’s assessment is more or less in accordance with the

model’s results for higher volatility rates. If volatility is, however, justiﬁably estimated less

pessimistically, then the bank —without further ado —may classify the loan one level better,

in line with its internal rating system, i.e. with 25 basic points credit risk. (The bank later

decided on the more optimistic evaluation of the position.)

11.3.6 Limitation of Loan Risk

The results in Table 11.4 show that the credit risk rises sharply if the time taken to realise the

ships were to draw further out or if the credit risk is considered over the entire term of the loan.

The more optimistic evaluation of the credit-worthiness of this ship mortgage can thus only be

justiﬁed if:

Applications 135

r

The credit-worthiness can be reassessed every three months, and if necessary the default

clause can be invoked.

r

The short time-scale for realising the ships, of from three to six months, is considered to be

realistic.

r

The volatility rate for the value of the ships is assessed, on the basis of a ﬁve-year time-scale –

in line with the residual term of the loan —on the optimistic side.

r

The shipowner’s management team is considered capable of operating the ships over the

coming ﬁve years in line with revenue expectations.

The above listing of preconditions, and of the relevant inﬂuencing factors involved, does on

the one hand show where the credit risks can be limited. It does, however, also become clear

that a bank requires much experience in the ﬁeld of ﬁnancing international shipping to be able

to operate this sort of business proﬁtably.

This example shows plainly that, thanks to the model, the credit risk may both be quanti-

ﬁed and also limited. It may also be demonstrated what risks the bank runs if the necessary

loan supervision measures are not undertaken. The model’s informative value is therefore

substantially higher than conventional methods of loan assessment.

11.4 MORTGAGE BUSINESS 1985–99

Over the period 1990–95 the banks in Switzerland were confronted with high losses on mort-

gage business. Our intention nowis to examine whether this situation could have been foreseen

with the aid of the model. The property index drawn up by the Cantonal Bank of Z¨ urich provides

the basis for the calculations (see Appendix 3).

Figure 11.1 gives a graphical representation of the course of the index. It makes it clear

that the index is at its most volatile for multiple dwelling units. The mortgage business is

50

1980 1985 1990

STE

EFH

MFH

1995 2000

100

I

n

d

e

x

Year

150

200

250

300

MFH is the abbreviation of the German text for multiple dwelling unit.

EFH is the abbreviation of the German text for single-family home.

STE is the abbreviation of the German text for condominium.

Figure 11.1

136 Risk-adjusted Lending Conditions

Table 11.5 Volatility of the index in %—for multiple dwelling units

as per Appendix 3

Year 4 Index values 5 Index values 6 Index values Maximum

1985 10.8 8.6 7.9 10.8

1986 10.7 8.5 7.4 10.7

1987 7.7 8.5 7.4 8.5

1988 1.5 6.7 7.4 7.4

1989 7.6 6.8 9.3 9.3

1990 24.2 19.2 16.4 24.2

1991 25.7 21.7 19.1 25.7

1992 6.9 20.2 18.7 2.2

1993 6.7 7.9 17.2 17.2

1994 2.2 5.5 7.3 7.3

1995 3.6 3.8 6.2 6.2

1996 21.3 17.6 15.0 21.3

1997 23.2 18.4 15.9 23.2

1998 20.5 18.2 15.6 20.5

1999 9.4 19.4 17.2 19.4

therefore investigated, using this index, in what follows. The course of the volatility is detailed

in Table 11.5. This was calculated for each calendar year with the help of the ﬁgures for the

last four, ﬁve and six year’s indices, such that three, four and ﬁve quotients respectively were

available for the calculation (see equation (7.45)).

The highest volatility ﬁgure concerned is given in each line in the last column of Table 11.5.

These ﬁgures are used later in the course of this analysis to deﬁne a worst-case scenario. This

means that both the inﬂuences both of short- and long-term volatility are included in the later

calculations. It is of course debatable whether this is the right way to proceed. For the purposes

of this example, however, the volatility in each case per annum is determined in this way. The

ﬁgures in the table are shown graphically in Figure 11.2. It is striking that the volatility ﬁgures

show a clear peak at the beginning and at the end of the 1990s. The cause is in the ﬁrst place

the sharp increase in the index ﬁgures and the subsequent sudden transition to a decline in

prices, and the uneven development in the second half of the 1990s.

Two assumptions vis-` a-vis mortgages are made for the credit shortfall risk calculation. The

ﬁrst assumption is that a multiple dwelling unit in 1985, i.e. well before the mortgage loan

crises at the beginning of the 1990s, is mortgaged at 80% of its saleable value. The amount of

the loan is left unchanged in subsequent years, without repayments, which leads to different

mortgage ﬁgures in each year corresponding to the course of the index. The credit shortfall

risk ﬁgures calculated in this way demonstrate how they developed for the bank over the years,

in line with the assumption that neither repayments of them, nor increases in them, took place.

The second assumption is that in each calendar year in each case a multiple dwelling unit

is mortgaged again to the extent of 80% of its saleable value. The credit shortfall risk ﬁgures

calculated in this way demonstrate what sort of risk the bank was running in each calendar

year concerned by ﬁnancing to the extent of 80%.

Applications 137

Table 11.6 Credit shortfall risk development in % under the

assumptions

Year 1985 1986 1987 1988 1989 1990 1991 1992

Ann. 1 0.09 0.00 0.00 0.00 0.00 0.01 0.10 0.02

Ann. 2 0.09 0.08 0.01 0.00 0.03 3.27 3.91 1.82

Year 1993 1994 1995 1996 1997 1998 1999 2000

Ann. 1 0.00 0.00 0.00 0.99 0.88 1.07 0.28

Ann. 2 0.98 0.00 0.00 2.18 2.85 1.90 1.57

Year

V

o

l

a

t

i

l

i

t

y

1985 1990 1995 2000

30%

25%

20%

15%

10%

5%

0%

4

5

6

Max

Figure 11.2

As may be inferred fromTable 11.6, the credit shortfall risks under the ﬁrst assumption were

very small even during the crisis at the beginning of the 1990s. Under the second assumption,

however, they became plainly higher than previously at the beginning of the 1990s. As early

as 1990 the ﬁgure was so high that 80% mortgages in that year were no longer appropriate.

This situation repeated itself towards the end of the 1990s. The facts are portrayed graphically

in Figure 11.3. Annual volatility ﬁgures were used for the calculation. Appropriately higher

ﬁgures should be used when granting ﬁxed mortgages over several years, which leads to credit

shortfall risks that are correspondingly even higher.

It becomes clear from Table 11.6 and Figure 11.3 that the mortgage loan crisis, based on the

annual index ﬁgures, was foreseeable from the model as early as the beginning of 1991, after

the index ﬁgure for 1990 came into existence. By using quarterly ﬁgures it may be assumed

that the reaction time could have been reduced even further.

It would be interesting, in the immediate future, to track down whether the credit shortfall

risks rates that have just risen again are leading to a second mortgage loan crisis in the near

138 Risk-adjusted Lending Conditions

1

2

4%

3%

2%

1%

0%

−1%

Year

1985 1990 1995 2000

C

r

e

d

i

t

s

h

o

r

t

f

a

l

l

r

i

s

k

Figure 11.3

future, or whether the necessary safety precautions, on the basis of the experiences banks have

had, have been taken.

On the strength of this result the usefulness of the model proves itself, at least in qualitative

terms. Whether or not the results are also sufﬁciently accurate in quantitative terms could only

be established by comprehensive empirical testing. Such testing would, however, far exceed

the scope of this study.

12

Final Considerations

So far we have emphasised that the method of calculating risk-adjusted loan interest rate

presented here is only a theory. Theories have to be examined empirically in order for their

validity to be corroborated (see Section 1.8). It was also pointed out that this empirical testing

would exceed the scope of these expositions. We will nonetheless outline in Section 12.1 how

this testing should be undertaken.

We will explain in Section 12.2 how a bank wishing to introduce the method described here

must proceed, and in Section 12.3 what the preconditions are for its applicability. We will at

the same time point out how closely the empirical testing, introduction and applicability are

connected with each other.

In Section 12.4 possible customer considerations will be discussed, and in Section 12.5

the questions still open will be summarised. We will conclude this study with some closing

remarks in Section 12.6.

12.1 TESTS OF HYPOTHESES

The method presented here is a theoretical, mathematical model for calculating the risks

involved for banks in their lending business. As the examples in the separate chapters show,

the results of the method are plausible as they match up in qualitative terms with the experience

of banks in this business. That does not, however, yet necessarily mean that the results are also

precise in quantitative terms. So it is necessary, in the next step, to test the method empirically

(see Section 1.8 too).

The construction of the empirical test is once again based on the concept of insurance. So

it has to be tested empirically, whether the ex ante forecasts losses agree with the ex post

actually established losses using the method presented here. But this is none other than the

comparison of imputed ‘insurance premiums’ with actual ‘loss experience’ (cf. Section 1.8;

[BCBS99, S. 50:a] backtesting). It follows from this that the empirical examination must be

built up by analogy with preliminary and actual costing in the insurance industry. And here

checks must be undertaken, using recognised statistical tests (cf. for example [BOHL92]) to

determine whether deviations between forecast and actual losses are coincidental, or whether

they follow a pattern.

Deviations that are coincidental lead to the conclusion that the theory does hold up. Devia-

tions that follow a pattern may have two causes:

r

that the theory does not hold up;

r

that the parameters selected do not hold up.

It is above all the determination of the ‘correct’ rates of volatility that is difﬁcult, as may

be inferred from subsection 7.4.2 and from the reading referred to there. In the cases of

140 Risk-adjusted Lending Conditions

deviations that follow a pattern it must therefore, above all, be checked whether any better

agreement between the forecast and the actual losses might have been achieved by selection of

‘better’ volatility rates. Then we must ﬁnd out how such ‘better’ volatility ﬁgures could have

been determined ex ante. Only if this approach is unsuccessful should the theory be rejected

(cf. Section 1.8 [BCBS99, S. 50: b) and c)]).

An independent examination of the method of procedure is also needed so that the model,

if required, becomes acceptable for regulatory purposes (cf. Section 1.8 [BCBS99, S. 50: d)]).

The expenditure required for the empirical testing outlined should not be underestimated in

any way at all. This becomes evident as one becomes aware of the prerequisites, which can be

subdivided into three main groups:

r

Creating the preconditions for the application of the method described here.

r

Creating the preconditions in operational accounting terms (also, possibly, with customers).

r

Creating the preconditions for the facts to be analysed.

Preconditions for the Application of the Method Described Here

These will be elaborated in the next section.

Preconditions in Operational Accounting Terms

In order to obtain the widest possible factual basis for statistical tests which in turn allows for

the most varied analyses, it is necessary from the start to record the following facts for each

loan utilised:

r

The agreed risk insurance premium r · L (see equation (1.1) for each interest period. De-

pending on whether or not the method is already being applied in the test phase on a 1/1

basis, either the risk premiums actually demanded of the customer are determined, or the

risk premiums calculated according to the model on the basis of the theory are obtained and

imputed.

r

The precautionary provisions actually made for each interest period, divided —for the pur-

pose of further analyses —into those required for regulatory purposes and those required

operationally for the banks’ own estimating purposes.

r

The losses actually incurred in each interest period.

r

The ﬁgures necessary for determining correlations in each interest period (see subsection

12.2.2 on ‘statistical bases’, subheading ‘correlations’).

r

The operational expenditure (personnel costs, etc.) incurred in processing winding up oper-

ations in each interest period. These costs do not in fact have anything directly to do with

the costs of risk discussed here, but they have to be covered too and, indeed, by proﬁt con-

tributions p · L (cf. Section 1.5, equation (1.1)). Even today there is probably not even one

bank that is aware of how high are the costs of processing just one single case of winding

up. If the operational accounting is already in place, then this too may be foreseen at the

same time.

Final Considerations 141

Clearly the expenditure required on IT for this is considerable, but it is nonetheless essential

for achieving our purposes. Simpliﬁcations may be made, if need be, at a later stage, if the theory

has proved well founded and when it has been possible to gather together the ﬁrst experiences

of its application by converting unique data into composite data, up to a certain point.

It is necessary to get right on top of the data ﬂow that the facts described above will form.

For this, in turn, the necessary IT resources have to be made available. On top of that banks will

have to recruit the necessary specialist staff who must be capable of designing and evaluating

statistical tests. Leading bank ofﬁcials, actuaries and economists familiar with the methods

described here will have to work together in such specialist teams. It must be assumed that

focused management of such a heterogeneous team will be critical.

The action described above will clearly be very expensive in time and money. The imple-

mentation of empirical testing must therefore be carefully planned, and budgeted for, in the

form of a project. The costs that are budgeted and the ongoing costs that will then arise have to

be set against the losses on loans that have been incurred in the past, and which have exceeded

the dimensions they were expected to have at the time. It is at this juncture that we venture to

forecast that this comparison will indeed demonstrate that the expenditure for such a project

will be justiﬁed.

12.2 IMPLEMENTATION IN PRACTICE

We can group the preconditions that must be fulﬁlled for the putting this method into practice,

as follows:

r

specialist personnel

r

statistical bases

r

IT support

r

organisational measures

12.2.1 Specialist Personnel

The necessary specialist personnel must be recruited for successful implementation of this

method, both in the ofﬁces that assess and wind up loans.

Loan Assessment Ofﬁces

The will must be there to recruit specialist personnel who have sufﬁcient ﬁnancial and mathe-

matical knowledge to be able to understand this theory. As becomes evident from the illustra-

tions in Section 7.5, the granting of loans for large ranges of mortgage and volatility ratings

is completely without problems. It is also the case that a majority of bank loans consists of

standard transactions such as advances against securities, ﬁnancing of owner-occupied houses,

and ﬁnancing of small companies on current account credit. It may be assumed that the great

majority of such business can be processed by standard, IT-supported procedures.

But specialists who are capable of making their own calculations are needed for assess-

ing more demanding loan transactions —whether it is more complex loan structures that are

142 Risk-adjusted Lending Conditions

concerned, or ﬁnancing arrangements that are more far-reaching. The results in these cases

in particular must be probed, and subjected to sensitivity analysis, for inaccuracies. A more

profound understanding of this theory is needed for such cases, even if much can be made

easier by IT support.

Winding Up Ofﬁces

The specialists in winding up ofﬁces must be capable, above all, of undertaking the calculations

described in Section 9.6. Some of this can most probably be standardised in terms of IT, in

cases of loan transactions with simple structures.

Sales

Sales staff must be capable of explaining the loan conditions that have been calculated to

customers. They must most of all be able to demonstrate, using illustrations, the correlations

between mortgage, volatility and loan interest. These play an especially important part in

advances against securities and the ﬁnancing of owner-occupied houses. The borrower against

securities is particularly concerned to prevent excess collateral for his loan being taken, while

the owner-occupier is particularly interested in obtaining the highest possible mortgage on his

house —especially when he is ﬁrst buying it. Here it must be demonstrated, with the aid of

tables and charts, that the level of mortgage in connection with volatility does have a direct

inﬂuence on the loan interest level. There is thus no sense in which there can be excess collateral

in the case of an advance against securities, i.e. the higher the collateral, the lower the rate of

interest, and vice versa. By way of contrast, there are quite clear upper limits when it comes to

the mortgaging of owner-occupied houses, where in the case of high mortgages interest rates

rise sharply such that, beyond a certain point, their acceptability is no longer to be assumed.

Similar considerations apply to the ﬁnancing of enterprises.

12.2.2 Statistical Bases

We have referred repeatedly to the importance of having available the necessary statisti-

cal bases. The ﬁgures for volatility (see subsection 7.4.2) are particularly critical in this

respect.

Financing Companies

In subsection 7.4.2 we made the assumption that, for determining the volatility of the market

value of a company, some three to four sets of annual accounts and one or two budgets for

the ﬁnancial years ahead must be available. Two questions arise here: on the one hand it must

be checked empirically, in accordance with the expositions in subsection 7.4.2, whether or

not the volatility was ascertained in this way with sufﬁcient precision. On the other hand,

borrowers must be in a position to deliver all this information. Regretfully must we record, at

this point, that these preconditions are not sufﬁciently fulﬁlled in the case of small and medium-

sized companies. One way of improving this situation lies in the opportunity it provides for

Final Considerations 143

calculating volatility rates on the basis of worst-case scenarios. The resulting high loan interest

might possibly motivate borrowers in future to prepare the necessary facts and ﬁgures, in order

to reduce their loan costs.

Mortgages

Reliable sequences of ﬁgures on the way prices have developed in the property market are

essential for being able to calculate mortgages reliably. The Cantonal Bank of Z¨ urich has

adopted a pioneering role here, in that it has demonstrated that this is possible [ZK96], with

the result that today it regularly publishes a property market index for the Canton of Z¨ urich

(see Appendix 3). Work in this ﬁeld must be carried further and extended to cover the whole

of Switzerland.

Loans Against Portfolios of Securities

The volatility of any portfolio of securities can be determined by reference to stock exchange

statistics. Here, too, empirical tests as described in Section 1.1 must demonstrate howsufﬁcient

reliability may be attained.

Correlations

The correlation between the probabilities of borrowers defaulting and of collateral falling short

was studied in Sections 8.2 and 8.3. Figure 8.1 illustrated that there must be differentiation

between four cases of the probabilities ˆ a,

ˆ

b, ˆ c, and

ˆ

d occurring for the determination of the

correlation. These ﬁgures must be regularly ascertained for every borrower of a covered loan,

in order to be able to calculate, based on that, the correlations that will be of interest.

12.2.3 IT Support

We have already mentioned, in Section 12.1, IT support for creating preconditions in terms

of operational accounting and in terms of statistical evaluation of facts. On top of that, the

IT-supported method described here must be implemented by means of expert systems. Due

attention must be paid to the numerical methods used, because of the complexity of the formu-

lae, and especially of the iterative calculations. Development of such expert systems should

therefore be left to software specialists in possession of estensive knowledge of numerical

systems. These would normally be software developers working on technical/scientiﬁc appli-

cations. No great volumes of facts actually have to be processed in the calculations relating to

any one single loan.

12.2.4 Organisational Measures

Putting this theory into practice calls for a comprehensive project, to be undertaken by staff of

appropriate calibre specially trained for it. This staff has to deal with the task areas, outlined

above, of staff recruitment and training, statistical bases and IT solutions.

144 Risk-adjusted Lending Conditions

Once the project workis concluded, a permanent specialist staff of mathematicians andstatis-

ticians must be set up, as companies in the insurance industry already know all too well. The

tasks for this staff include development of the theory, handling complex special cases, keeping

the statistical bases permanently up to date, and ongoing development of the IT-based aids.

12.3 APPLICABILITY OF THE METHOD PRESENTED

To be able to apply the method presented, the following preconditions must be fulﬁlled:

r

The theory must be corroborated empirically, as described in Section 12.1.

r

The operational preconditions as described in Section 12.2 must be fulﬁlled.

The crux of the matter here is the following: the theory must already be introduced de facto in

order for it to be possible to test it empirically. This was indeed pointed out in Section 12.1.

Here it is not so important, from the point of view of expenditure, whether the introduction

is at ﬁrst made partly by imputation, or wholly in transactions with customers. If a bank does

decide to undertake empirical testing, then it must in actual fact be taking a de facto decision

to bring the method in.

One wayof reducingcosts couldlie inrestrictingimplementationtojust one part of the bank’s

operations (for example to one type of loan transaction, or within one limited geographical

area).

That way costs, particularly for specialist staff and if need be for the working up of statistical

bases, could be limited. The costs of IT and of the project team would, however, remain more

or less unchanged.

The question then arises: howbig is the danger of failure and what does it cost? The results of

the theory are plausible in qualitative terms, as has already been mentioned at the beginning of

Section 12.1. Many other examples that are not set down here have conﬁrmed this impression.

So it is permissible at this juncture to assume that absolute failure is ruled out. So the risk

is rather that the method may not deliver as much as it seems to promise. This, however, is

not least a question of expectations. It remains therefore a matter of weighing costs against

beneﬁts. We make no secret of the fact that banks will have to reckon with considerable costs.

On the other hand, however, the possible beneﬁts of really getting the costs of credit risks

under control, are likewise considerable.

Implementation of this method would presumably be at its most simple for a bank entering

into the business of lending for the ﬁrst time. As such a bank would have to be building up the

necessary IT structures fromscratch, anyway, it might do so as suggested here. In order to limit

the risk of mistakes arising from the theory, applications for loans should still be examined in

the conventional manner and then checked by the method presented here. Loans should then

only be granted if both methods of operation come to a positive result, and on the basis that it

should be the higher rate of loan interest that is applied to the commercial transaction with the

customer. This does not jeopardise the empirical test in any way at all: under the theory put

forward here, interest rates that are higher than the minimum rates are certainly permissible

(cf. Section 4.6). But in these cases the distinction must be drawn in operational accounting

between risk premiums necessary according to the model and risk premiums that are in practice

charged and collected.

Final Considerations 145

This section concludes with a strong warning against making empirical testing —out of cost

considerations —no more than a ‘trawl’ through the archives. Our own experiences with such

exercises show that (a) they do not as a rule come up with the desired results, and (b) they do

not really save any money.

One must not underestimate, either, the time it takes for reliable facts to accumulate. The

Basel Committee itself reckoned on several years (cf. Section 1.8: [BCBS99, S. 2]). According

to the terms of the loan products to be examined, it is assumed here that between about ﬁve

and 10 years must be allowed. But it should also be taken into account that know-how will be

building up continually —from ongoing evaluations —until the situation we are aiming for is

reached.

12.4 CUSTOMER CONSIDERATIONS

As already mentioned in Section 10.10, something is both demanded and expected of the

borrower when this model is applied. It should not therefore be ruled out that a bank customer

might not be prepared to make the efforts demanded of it. As we explained in Section 1.3, the

borrower decides, in the same way as the concept of insurance is applied, whether or not it

is prepared to pay the risk-adjusted price being asked. It will not necessarily do this if more

favourable counter offers are available. The bank has to make the borrower aware, in such

situations, why it has reached the conclusions it has reached. The bank has to decide, on the

strength of an assessment of the position —in which the overall current and future business

potential of this customer and the credit risks involved with it are weighed against each other —

whether or not it is willing to persevere with its offer and its demands for information. If it is

sure that it has calculated the loan interest rate correctly, it will do this in its own interest, as

in the last analysis it jeopardises its own existence if it makes too many concessions in respect

of its price and demands for information.

It is at this point that we venture to forecast that the banks that will be successful in the long

term, in the business of lending, will be those which can indeed say no —when their clients

obtain more favourable counter offers from other banks —provided that they themselves are

in a position to calculate risk-adjusted loan conditions correctly. From this it will follow that

the borrower will have to reﬂect whether it always wants to go for the cheapest offer —

with the disadvantage that this may be associated with frequent changes of bank partners —or

whether it is on occasion prepared, in the interest of a longer-termconﬁdence-building business

relationship, to meet the bank’s information requirements and to pay the price the bank has

asked.

12.5 OPEN QUESTIONS

As we mentioned in Section 1.2, the subject of this study was deliberately limited. In particular

we did not cover the following in terms of separate business transactions:

r

borrowers in the public sector and/or ratings related to countries

r

loans in foreign currencies

r

international lending business, especially the ﬁnancing of projects.

146 Risk-adjusted Lending Conditions

Furthermore the subject of portfolios of loans was likewise not covered. Efforts are being made

to develop methods that scale down the shortfall risk of a whole portfolio of loans vis-` a-vis the

sum of the individual loan risks (cf. for example, [MANZ98] and the reading suggested there).

To what extent the method we have introduced here may contribute anything in this ﬁeld must

be the subject of future studies.

12.6 CLOSING REMARKS

Some idea of the consequences of the method we have introduced here may have been com-

municated with the help of illustrations and examples, but it has been necessary to conﬁne

ourselves to a small selection. The meaningful content of this theory has thus not yet been

fully developed by any means, and the ﬁeld for further studies is therefore still wide open.

In portraying the method we have introduced, it has only been possible to give an overall

view. Many details still await clariﬁcation.

So it is as true as ever that providing the answers to old questions all too often forms the

starting point for posing new ones!

Appendix 1: Notation

Equation number

ˆ a probability 8.2

b breakdown distribution rate probability 2.4

ˆ

b probability 8.2

b

c

corrected breakdown distribution rate probability 7.56

ˆ c probability 8.2

d debt rate 7.30

ˆ

d probability 8.2

e return on equity 9.22

f ﬁnancing cost rate 1.1

g return rate 9.21

i loan interest rate 1.1

i (t ) loan interest rate according to the whole term 7.50

i

c

(t ) corrected loan interest rate according to the whole term 7.56

i

d

discount rate 7.41

i

g

return on a risk-free investment in government bonds 7.43

i

m

marginal interest rate 9.1

i

mt

return on the market 7.43

i

pa

interest rate on annual basis 7.40

i

s

risk-free standard interest rate 1.2

i

spa

annual standard interest rate 7.40

p proﬁt contribution rate 1.1

r shortfall risk hedging rate 1.1

r

c

correction factor 4.21

r

cac

current account credit write-off risk hedging rate 4.20

ˆ r correlation coefﬁcient 8.3

t term 7.37

v return on assets 9.21

A number of trading days per year 8.1

B breakdown distribution probability value 2.4

148 Risk-adjusted Lending Conditions

B

c

corrected breakdown distribution 7.54

C free cash ﬂow 7.41

C

j

cash ﬂow face value after j periods 1.2

D debts 7.2

E equity 7.1

L payed out loan amount 1.2

L

g

granted loan 4.19

L

u

used loan 4.20

N standard normal distribution function 7.28

P put value 7.4

R

b

billed return 4.20

R

n

necessary return 4.19

S proportional salaries 7.54

V value of the company 7.1

V

I

liquidation value 7.46

X portfolio values 8.1

β measurement of unlevered assets market risk according to CAPM 7.43

χ survival chance 2.2

χ

∗

survival chance regarding the breakdown distribution probability value rate 2.8

ε cumulative success chance 3.2

ϕ cumulative shortfall risk 3.2

x credit-worthiness key ﬁgure 2.10

µ medium of the logarithms 7.45

ρ shortfall risk 2.1

ρ

∗

credit shortfall risk 2.7

ρ

a

average of all n shortfall risks ρ

k

5.8

ρ

c

corrected credit risk 7.57

ρ

B ∩C

combined risk 8.2

σ volatility of the company value according to the time period 7.28

ψ

j

probability of cash ﬂow C

j

1.2

Γ gamma function 7.45

Λ loan market value 4.1

Appendix 2: Excel Worksheet

An Excel Worksheet set up as follows proved to be the best way of calculating the

equations in Chapter 7. [For this iterations absolutely must be allowed under Extras/

Options/Calculations.]

A B

1 Debts L

tot

(Input ﬁeld)

2 Amount of loan L (Input ﬁeld)

3 Term t in years (Input ﬁeld)

4 Standard interest rate i

s

(Input ﬁeld)

5 Market value of the company V (Input ﬁeld)

6 Volatility of the market value σ (Input ﬁeld)

7 Privileged salary/wage claims S (Input ﬁeld)

8 Mortgage d = B1/B5

9 Volatility according to term t = B6

∗

SQRT(B3)

10 x = LN(B8/(1-B13))/B9+B9/2

11 N(x −σ(t )) = STANDNORMDIST(B10-B9)

12 N(x): probability of bankruptcy ρ = STANDNORMDIST(B10)

13 Loan risk ρ

∗

= (B11-B8

∗

B12)/(B11-B8)

14 Loan risk per annum ρ

pa

1-(1-B13)ˆ(1/B3)

(Continued)

150 Risk-adjusted Lending Conditions

15 Loan interest rate i

pa

(B4+B11)/(1-B14)

16 Breakdown distribution rate b 1-B13/B12

17 Loan demand at maturity B2

∗

(1+B15)ˆB3

18 Breakdown distribution probability

value B B16

∗

B17

19 Proportional privileged salary/wages

claims S B7

∗

B2/B1

20 Corrected value of B: B

c

IF(B18>B19;B18-B19;0)

21 Corrected breakdown distribution rate B20

∗

probability b

c

(1-B12)/(B2

∗

(1+B4)ˆB3-B20

∗

B12)

22 Corrected loan risk ρ

∗

c

B12

∗

(1-B21)

23 Corrected loan risk per annum ρ

∗

cpa

1-(1-B22ˆ(1/B3)

24 Corrected loan interest rate i

cpa

(B4+B23)/(1-B23)

25 Corrected loan interest rate on account of S B24-B15

Hint 1: The loan interest rates in B15 and B24 are calculated on the basis of the loan risks

calculated in B14 and B23, and not on the basis of the shortfall risk of the rating level

concerned according to Table 2.1. Table 2.1 may indeed be programmed in Excel,

but this is very laborious.

Hint 2: On occasion Excel manifests difﬁculties with this worksheet and reacts with various

fault reports in the B column. If this happens click on Field B13, and then F2, and

then press ENTER.

Hint 3: In the case of unreliable ﬁgures for d (Field B7): i.e. d ≤ 0 or d ≥ 1; Excel runs

through the maximum number of iterations (presetting usually 100; can be seen

under EXTRAS/OPTIONS/CALCULATIONS) and then aborts with fault reports.

Reducing the maximum number of iterations under the presetting is therefore rec-

ommended. Usually 10 will sufﬁce.

Hint 4: If only the loan amount in line 2 is changed from one calculation to the next, then

only the values in lines 21 to 25 change in the results.

Appendix 3: Property Price Index

The Cantonal Bank of Z¨ urich’s property price index for the Canton of Z¨ urich, dated 5 May

2000.

Year Single-family home Multiple dwelling unit Condominium

1980 100.0 100.0 100.0

1981 115.0 102.9 103.1

1982 124.1 110.4 113.5

1983 132.0 120.3 110.6

1984 138.0 147.1 119.5

1985 143.5 148.8 124.6

1986 150.7 167.6 129.1

1987 167.0 192.2 144.9

1988 200.5 222.3 161.0

1989 223.2 287.5 179.6

1990 243.6 245.7 193.6

1991 225.8 219.5 189.7

1992 226.0 211.8 196.2

1993 214.6 212.6 188.9

1994 221.9 208.9 193.1

1995 216.3 218.7 183.9

1996 200.2 160.6 178.3

1997 192.9 168.4 171.4

1998 188.0 156.9 161.8

1999 190.2 171.6 162.7

http://www.zkb.ch/bin/entry/frame/private/immobilien/index.html

Source: Cantonal Bank of Z¨ urich

Appendix 4: Chapter 3 —Derivations

First of all we calculate the values of ϕ

j

and ε

j

for each individual period of the n periods.

The solution for the ﬁrst period is evident:

ϕ

1

= ρ

1

(3.3)

ε

1

= χ

1

(3.4)

It is assumed, for further considerations, that ρ

j +1

and χ

j +1

are independent of ρ

j

and χ

j

.

For a borrower deﬁnitely either to default or not to default in the second year, it may not default

in the ﬁrst year. According to the laws of multiplication of probability calculus [BOHL92, S.

324] the following applies:

ϕ

2

= χ

1

· ρ

2

(3.5)

ε

2

= χ

1

· χ

2

(3.6)

The same applies, by analogy, for the following periods:

ϕ

j, j =1

=

j −1

¸

k=1

χ

k

· ρ

j

=

j −1

¸

k=1

(1 −ρ

k

)

· ρ

j

(3.7)

ε

j

=

j

¸

k=1

χ

k

(3.8)

On the assumption that ρ

j

and χ

j

have the same value, in each case, for all n periods, equations

(3.7) and (3.8) may be simpliﬁed as follows:

ϕ

j

= χ

( j −1)

· ρ = (1 −ρ)

( j −1)

· ρ if ρ

1

= · · · = ρ

n

= ρ (3.9)

ε

j

= χ

j

if χ

j

= · · · = χ

n

= χ (3.10)

Under the deﬁnition above ϕ(n) means the probability that the borrower will default not in

any speciﬁed but in any one of the n periods. As the default may only occur in one and not in

several periods, if it occurs at all, the following self-excluding events generally [BOHL92, S.

154 Risk-adjusted Lending Conditions

313] apply according to the laws of addition of probability calculus:

ϕ(n) =

j

¸

j =1

ϕ

j

= ρ

1

+

n

¸

j =2

¸

j −1

¸

k=1

(1 −ρ

k

)

· ρ

j

(3.11)

and, especially:

ϕ(n) =

n

¸

j =1

(1 −ρ)

( j −1)

· ρ if ρ

1

= · · · = ρ

n

= ρ (3.12)

According to equation (3.12) ϕ(n) is a geometric series with a

1

= ρ as its ﬁrst term and the

multiplication factor q = (1 −ρ), under which the following applies:

ϕ(n) = ρ ·

1 −(1 −ρ)

n

1 −(1 −ρ)

= 1 −(1 −ρ)

n

= 1 −χ

n

(3.13)

The law of addition does not apply to ε(n), as in contrast to shortfall, success only occurs

if each individual period of the n periods was successful. ε(n) thus corresponds to ε

n

(cf.

equations (3.8) and (3.10)). So in general the following applies:

ε(n) =

n

¸

j =1

χ

j

(3.14)

and, especially:

ε(n) = χ

n

= (1 −ρ)

n

if χ

1

= · · · = χ

n

= χ (3.15)

The following thus applies in the special case:

ϕ(n) +ε(n) = (1 −χ

n

) +χ

n

= 1 (3.16)

which equation (3.2) conﬁrms. In the general case the following results:

ϕ(n) = 1 −

n

¸

j =1

χ

j

= 1 −

n

¸

j =1

(1 −ρ

j

) (3.17)

Reﬂection now leads to the probability ψ

j

, that ψ

j

is none other than the probability that

no loss arises within the ﬁrst j periods. So in general the following applies:

ψ

j

= ε

j

=

j

¸

k=1

χ

k

=

j

¸

k=1

(1 −ρ

k

) (3.18)

and, especially:

ψ

j

= χ

j

= (1 −ρ)

j

if χ

1

= · · · = χ

n

= χ (3.19)

Appendix 5: Chapter 4 —Derivations

SECTION 4.1 DERIVATION

According to the assumptions that were made (cf. Section 1.7), the same risk-free rate of

interest for discounting is used for each summand. Furthermore, Λ = L applies. Reducing

with L and putting i outside the brackets in the ﬁrst summand results in:

1 = i ·

n

¸

j =1

χ

j

(1 +i

s

)

j

+

χ

n

(1 +i

s

)

n

+

n

¸

j =1

χ

j −1

· ρ · b · 1 +i

1 +i

j

s

(4.2)

In the case of the ﬁrst summand, we are concerned with a geometric series with:

a

1

= ρ =

χ

(1 +i

s

)

(4.3)

In the case of the second summand, we are likewise concerned with a geometric series with:

a

1

=

ρ · b · (1 +i )

(1 +i

s

)

ρ =

χ

(1 +i

s

)

(4.4)

So ρ is identical in both geometric series and only a

1

is different.

Byusingthe overall equationfor geometric series andsubstitutionbyequation(4.3), equation

(4.2) is rewritten as follows:

1 =

i · ρ · (1 −ρ

n

)

(1 −ρ)

+ρ

n

+

ρ · b · (1 +i ) · (1 −ρ

n

)

(1 +i

s

) · (1 −ρ)

(4.5)

Taking the mean summand ρ

n

on the left-hand side and extending by (1 −ρ) results in:

(1 −ρ) · (1 −ρ

n

) = i · ρ · (1 −ρ

n

) +

ρ · b · (1 +i )

(1 +i

s

)

· (1 −ρ

n

) (4.6)

The fact that the equation may be shortened at this point by (1 −ρ

n

), is important. This now

results in:

(1 −ρ) = i · ρ +

ρ · b · (1 +i )

(1 +i

s

)

(4.7)

156 Risk-adjusted Lending Conditions

So the path of the term with the time element of the number of periods n has been shortened

as a consequence of the assumption that shortfall risks are constant. The key result for the

shortfall risk hedging rate r comes thus to be independent of the term of the loan! Tasking i

out of brackets results in:

(1 −ρ) = i ·

ρ +

ρ · b

(1 +i

s

)

+

ρ · b

(1 +i

s

)

(4.8)

Reverse substitution of ρ and replacement of i by (i

s

+r) according to equations (1.1) and

(1.3) results in:

1 −

χ

(1 +i

s

)

= (i

s

+r) ·

χ +ρ · b

1 +i

s

+

ρ · b

(1 +i

s

)

(4.9)

Multiplication by (i

s

+r) and replacing χ by (1 −ρ) results in:

i

s

+ρ = (i

s

+r) · (1 −ρ +ρ · b) +ρ · b (4.10)

Multiplying out and insertion of (1 −ρ +ρ · b) = (1 −ρ

∗

) (see equation (2.6)) results in:

i

s

+ρ = i

s

· (1 −ρ

∗

) +r · (1 −ρ

∗

) +ρ · b (4.11)

Solution by r results in:

r =

i

s

+ρ −i

s

· (1 −ρ

∗

) −(ρ · b)

(1 −ρ

∗

)

(4.12)

r =

i

s

+ρ

∗

−i

s

+i

s

· ρ

∗

(1 −ρ

∗

)

(4.13)

r =

ρ

∗

1 −ρ

∗

· (1 +i

s

) (4.14)

As r is independent of n, the same shortfall risk hedging rate r must also be valid for a

limitless number of periods of loan term, in which the expectation value of the repayment of

capital according to Section 3.3.1 is precisely zero! This can be veriﬁed, in that equation (4.5)

is written for a limitless geometric series with a limitless number of periods, thus:

1 = i ·

ρ

1 −ρ

+ρ

∞

+

ρ · b(1 +i )

(1 +i

s

) · (1 −ρ)

(4.15)

As χ ≤ 1 always applies, ρ < 1 always applies on the assumption that i

s

> 0 always applies

likewise. This assumption is permissible, as there is in practice no such thing as ‘free credit’.

This means that ρ

∞

is a null consequence and equation (4.15) turns into:

1 = i ·

ρ

1 −ρ

+

ρ · b(1 +i )

(1 +i ) · (1 −ρ)

(4.16)

(1 −ρ) = i · ρ +

ρ · b · (1 +i )

(1 +i

s

)

(4.17)

Appendix 5: Chapter 4 —Derivations 157

The comparison shows that equations (4.7) and (4.17) are identical and thus must lead to the

same conclusion. The value

ρ

∗

= ρ(1 −b) (4.18)

is none other than the credit shortfall risk, which is dependent on the shortfall risk ρ of the

borrower and on the probable breakdown distribution rate b.

SECTION 4.5 DERIVATION

The deviation L from the nominal amount is calculated as the difference of:

L = Λ − L (4.30)

This means:

L > 0 appreciation proﬁt

L < 0 need for provision to be made

From (4.29) and (4.30) there results:

L = Λ − L =

l

¸

j =1

χ

j

l

· i · L

(1 +i

sl

)

j

+

χ

l

l

· L

(1 +i

sl

)

l

+

l

¸

j =1

χ

j −1

l

· ρ · b · L · (1 +i )

(1 +i

sl

)

j

− L

(4.31)

Abbreviation using L results in:

L

L

= λ =

l

¸

j =1

χ

j

l

· i

(1 +i

sl

)

j

+

χ

l

l

(1 +i

sl

)

l

+

l

¸

j =1

χ

j −1

l

· ρ

l

· b · (1 +i )

(1 +i

sl

)

j

−1 (4.32)

In the case of both summands, we are again concerned with geometric series. By using the

appropriate overall equations, there results:

λ =

i ·

χ

l

(1 +i

sl

)

·

1 −

χ

l

l

(1 +i

sl

)

l

1 −

χ

l

(1 +i

sl

)

+

χ

l

l

(1 + i

sl

)

l

+

ρ

l

· b · (1 +i ) ·

1 −

χ

l

l

(1 +i

sl

)

l

(1 + i

sl

) ·

1 −

χ

l

(1 +i

sl

)

−1 (4.33)

Reformulation of the ﬁrst and third summands gives:

λ =

i · χ

l

·

1 −

χ

l

l

(1 +i

sl

)

l

(1 + i

sl

−χ

l

)

+

χ

l

l

(1 + i

sl

)

l

+

ρ

l

· b · (1 +i ) ·

1 −

χ

l

l

(1 +i

sl

)

l

(1 + i

sl

−χ

l

)

−1 (4.34)

Giving them a common denominator results in:

λ =

i · χ

l

· (1 + i

sl

)

l

·

1 −

χ

l

l

(1 +i

sl

)

l

+χ

l

l

· (1 + i

sl

−χ

l

)

(1 + i

sl

−χ

l

) · (1 + i

sl

)

l

(4.35)

+

ρ

l

· b · (1 +i ) · (1 + i

sl

)

l

·

1 −

χ

l

l

(1 +i

sl

)

l

−(1 + i

sl

−χ

l

) · (1 + i

sl

)

l

(1 + i

sl

−χ

l

) · (1 + i

sl

)

l

158 Risk-adjusted Lending Conditions

Multiplying them out in the numerator results in:

λ =

i · χ

l

·

¸

1 + i

l

sl

−χ

l

l

¸

+χ

l

l

· (1 + i

sl

−χ

l

)

(1 + i

sl

−χ

l

) · (1 + i

sl

)

l

+

ρ

l

· b · (1 + i ) ·

¸

1 + i

l

sl

−χ

l

l

¸

−(1 + i

sl

−χ

l

) · (1 + i

sl

)

l

(1 + i

sl

−χ

l

) · (1 + i

sl

)

l

(4.36)

Taking (1 +i

sl

)

l

and χ

l

l

out of brackets gives:

λ =

(1 + i

sl

)

l

· (i · χ

l

+ρ

l

· b · (1 + i ) −1 −i

sl

+χ

l

)

(1 + i

sl

−χ

l

) · (1 + i

sl

)

l

+

χ

l

l

· (1 + i

sl

−χ

l

−i · χ

l

−ρ

l

· b · (1 + i ))

(1 + i

sl

−χ

l

) · (1 + i

sl

)

l

(4.37)

The two right-hand brackets in the summands of the numerator are identical up to the reversed

plus/minus sign, and may thus be taken out of brackets. In addition, other terms in these

brackets may be put together, so that the following conclusion emerges:

λ =

¸

1 + i

l

sl

−χ

l

l

¸

· [(1 +i ) · (ρ

l

· b +χ

l

) −(i

sl

+1)]

(1 + i

sl

)

l

· (1 + i

sl

−χ

l

)

(4.38)

According to equation (4.32), − is the valuation correction in percentage terms in relation to

the nominal value of the loan.

Appendix 6: Chapter 5 — Derivations

SECTION 5.2 DERIVATION

First of all equation (5.10) should be simpliﬁed for y. By applying the approximation formula

[DMK/DPK92, S. 50]:

(1 + x)

a

≈ 1 +ax (5.12)

can be written:

y ≈

1 −nρ

a

−

n

¸

k=1

ρ

k

(1 +i

s

)

n

(5.13)

The summand in the numerator is by deﬁnition equal to zero, as what we are concerned with

here is the total of the deviations from the mean:

n

¸

k=1

ρ

k

= 0 (5.14)

So one obtains:

y ≈

1 +nρ

a

(1 +i

s

)

n

(5.15)

Renewed application of the approximation equation (5.12) in the numerator of formula (5.15)

leads to:

y ≈

1 −ρ

a

1 +i

s

n

(5.16)

On the assumption that it is, by way of approximation, permissible to use this approximate

solution also for only j periods instead of for all n periods, equation (5.9) can be written as

160 Risk-adjusted Lending Conditions

follows:

x ≈

n

¸

j =1

1 −ρ

a

1 +i

s

j

(5.17)

Here it is presupposed that for all values of j

j

¸

k=1

ρ

k

≈ 0 (5.18)

applies with sufﬁcient precision.

The sum in equation (5.17) is a geometric sequence with:

a

1

= ρ =

1 −ρ

a

1 +i

s

(5.19)

This leads to the following result:

x ≈

ρ · (1 −ρ

n

)

1 −ρ

(5.20)

By use of the substitution (5.19) equation (5.16) can be written as follows:

y ≈ ρ

n

(5.21)

By analogy with equation (5.17), equation (5.11) can, by way of approximation, be written as

follows:

z ≈

n

¸

j =1

(ρ

a

+ρ

j

) · (1 −ρ

a

)

j −1

(1 +i

s

)

j

=

n

¸

j =1

(ρ

a

+ρ

j

)

(1 −ρ

a

)

·

1 −ρ

a

1 +i

s

j

(5.22)

On the assumption that ρ

j

may, when adding up, be ignored — also by way of approxima-

tion — and by using equations (5.17) and (5.19) equation (5.22) can be written as follows:

z ≈

ρ

a

1 −ρ

a

·

ρ · (1 −ρ

n

)

1 −ρ

(5.23)

The insertion of equations (5.20), (5.21) and (5.23) into equation (5.6) results in:

i ≈

1 −ρ

n

−

b · ρ

a

· ρ · (1−ρ

n

)

(1−ρ

a

) · (1−ρ)

ρ · (1−ρ

n

)

(1−ρ)

+

ρ · (1−ρ

n

)

(1−ρ

a

) · (1−ρ)

(5.24)

Reduction by (1 −ρ

n

) and extension by (1 −ρ) results in:

i ≈

1 −ρ −

b · ρ

a

· ρ

(1−ρ

a

)

ρ +

b · ρ

a

· ρ

(1−ρ

a

)

=

1 −ρ ·

1 +

b · ρ

a

1−ρ

a

ρ ·

1 +

b · ρ

a

1−ρ

a

(5.25)

Appendix 6: Chapter 5 — Derivations 161

Reverse substitution gives:

i ≈

1 −

1 −ρ

a

1 +i

s

·

1 +

b · ρ

a

1 −ρ

a

1 −ρ

a

1 +i

s

·

1 +

b · ρ

a

1 −ρ

a

(5.26)

Multiplying out gives:

i ≈

1 −

1 −ρ

a

+b · ρ

a

(1 +i

s

)

1 −ρ

a

+b · ρ

a

(1 +i

s

)

(5.27)

Extension by (1 +i

s

) and taking ρ

a

out of brackets gives:

i ≈

1 +i

s

−1 +ρ

a

· (1 −b)

1 −ρ

a

· (1 −b)

=

i

s

+ρ

a

· (1 −b)

1 −ρ

a

· (1 −b)

(5.28)

i ≈

i

s

+ρ

∗

a

1 −ρ

∗

a

(5.29)

Comparison with equation (4.49) reveals that this is identical to equation (5.29). In the case of

the variable ρ

j

, equation (4.49) may thus — by way of approximation — likewise be used, in

that the average ρ

a

replaces the constant ρ.

SECTION 5.4 DERIVATION

Equation (5.30) may now, owing to the symmetries, be summarised as follows:

L =

n

¸

j =1

¸¸

i · L +

b · ρ

j

· (1 +i )

(1−ρ

j

)

· L

¸

·

¸

1 −

( j −1)

n

¸

+

L

n

¸

·

j

¸

k=1

(1 −ρ

k

)

(1 +i

s

)

j

(5.31)

After abbreviation by L, only the curved brackets are reformulated at ﬁrst:

{” ”} =

¸

i +

b · ρ

j

1 −ρ

j

+i ·

b · ρ

j

1 −ρ

j

¸

·

¸

n − j +1

n

¸

+

1

n

(5.32)

{” ”} =

¸

i ·

1 +

b · ρ

j

1 −ρ

j

+

b · ρ

j

(1 −b

j

)

¸

· [n − j +1] +1

n

(5.33)

{” ”} =

i ·

1 +

b·ρ

j

1−ρ

j

· (n − j +1) +

b · ρ

j

· (n−j +1)

1−ρ

j

+1

n

(5.34)

{” ”} =

i

n

·

¸

1 +

b · ρ

j

1 −ρ

j

+

b · ρ

j

(1 +ρ

j

) · i

· (n − j +1) +

1

i

¸

(5.35)

{” ”} =

i

n

·

¸

1 +

b · ρ

j

1 −ρ

j

·

1 +

1

i

· (n − j +1) +

1

i

¸

(5.36)

162 Risk-adjusted Lending Conditions

Inserted into equation (5.30) abbreviated by L, and reformulated results in:

n

i

=

n

¸

j =1

¸

1 +

b · ρ

j

1 −ρ

j

·

1 +

1

i

· (n − j +1) +

1

i

¸

·

j

¸

k=1

(1 −ρ

k

)

(1 +i

s

)

j

(5.37)

and transposed into:

i =

n

n

¸

j =1

¸

1 +

b·ρ

j

1−ρ

j

·

1 +

1

i

· (n − j +1) +

1

i

¸

·

j

¸

k=1

(1 −ρ

k

)

(1 +i

s

)

j

(5.38)

Bibliography

ALT89 Altman, E. I.: Measuring Corporate Bond Mortality and Performance; in: Journal of Finance,

44; 1989.

ALTM77 Altmann, E. I.; Haldeman, R. G.; Narayana, P.: Zeta Analysis: a New Model to Identify

Bankrupty Risk of Corporations; in: Journal of Banking and Finance, June 1977, 29–54.

AUCK94 Auckenthaler, Christoph: Finanzmathematische Grundlagen des Investment Banking; Bern,

Stuttgart, Wien 1994.

BCBS99 Basle Committee on Banking Supervision: Credit Risk Modelling: Current Practices and

Applications; Basel 1999 (http://www.bis.org/wnew.htm).

BCCH97 Bakshi, Gurdip; Cao, Charles; Cen, Zhiwu: Empirical Performance of Alternative Option

Pricing Models; in: Journal of Finance, 52 (2), December 1997.

BEDE97 Berger, Allen N.; DeYoung, Robert: Problem Loans and Cost Efﬁciency in Commercial

Banks; in: Journal of Banking & Finance, 21; 1997.

BLSC73 Black, Fischer; Scholes, Myron: The Pricing of Options and Corporate Liabilities; in: Journal

of Political Economy; May–June 1973.

BOES96 Boeschenstein, Roland: Der Firmenkunde will mehr; in: Handelszeitung Nr. 38, 19.09.1996;

Z¨ urich 1996.

BOHL92 Bohley, Peter: Statistik, Einf ¨ uhrendes Lehrbuch f ¨ ur Wirtschafts- und Sozialwissenschaftler,

5.Auﬂage; M¨ unchen, Wien, Oldenburg 1992.

BRAK91 Brakensiek, Thomas: Die Kalkulation und Steuerung von Ausfallrisiken im Kreditgesch¨ aft

der Banken; Frankfurt am Main 1991.

BRMY96 Brealey, Richard A.; Myers, Stewart C.: Principles of Corporate Finance; ﬁfth edition; New

York 1996.

BRUN94 Brunner, Christoph: Bank¨ ubernahmen in der Schweiz; Bern, Stuttgart, Wien 1994.

BRVA97 Briys, Eric; Varenne de, Fran¸ cois: Valuing Risky Fixed Rate Debt: An Extension; in: Journal

of Financial and Quantitative Analysis, 32 (2); June 1997.

B

¨

UEV B¨ uschgen, Hans E.; Everling, Oliver (Herausgeber): Handbuch Rating; Wiesbaden 1996.

CAFA92 Carlson, John H.; Fabozzi, Frank J.: The Trading and Securitization of Senior Bank Loans;

Chicago, Cambridge 1992.

CAOU98 Caouette, J. B.; Altmann, E. I.; Narayana, P.: Managing Credit Risk: the Next Great Financial

Challenge; New York 1998.

CART98 Carty, Lea V.: Credit Risk and Credit Quality Correlation: What Can We Measure?; in: Journal

of Lending & Credit Risk Managment; February 1998.

CHO98 Cho, Myeong-Hyeon: Ownership Structure, Investment, and Corporate Value: an Empirical

Analysis; in: Journal of Financial Economics, 47 (1), January 1998, S. 103–121.

CHRI97 Chriss, Neil A.: Black–Scholes and Beyond —Option Pricing Models; Chicago/

London/Singapore 1997.

164 Risk-adjusted Lending Conditions

COME91 Cooper, I.; Mello, A.: The Default Risk of Swaps; in: Journal of Finance, 46; 1991.

COPI01 Cossin, D.; Pirotte, H.: Advanced Credit Risk Analysis; Chichester 2001.

CORU85 Cox, John C.; Rubinstein, Mark: Options Markets; Englewood Cliffs 1985.

CPSH98 Carey, Mark; Part, Mitch; Sharp, Stevan A: Does Corporate Lending by Banks and

Finance Companies Differ? Evidence on Specialization in Private Debt Contracting;

in: Journal of Finance, 53, June 1998 (http://www.cob.ohio-state.edn/htbin/htimage/˜ﬁn/

journal/jf.conf? 94,154).

DICH98 Dicher, IIia D.: Is the Risk of Bankruptcy a Systematic Risk?; in: Journal of Finance, 53,

June 1998 (http://www.cob.ohio-state.edn/htbin/htimage/˜ﬁn/journal/jf.conf?94,154).

DMK/DPK92 Deutschschweizerische Mathematikkommission; Deutschschweizerische Physikkommis-

sion: Formeln und Tafeln Mathematik —Physik, 5. Auﬂage; Z¨ urich 1992.

DOSA97 Domowitz, la; Sartain, Robert L.: Determinants of the Consumer Bankruptcy Decision;

in: Journal of Finance, forthcoming, http://www.cob.ohiostate-edu/htbin/htimage/˜ﬁn/

journal/jf.conf?94,154.

DRZI98 Drzik, John: The Seven Stages of Risk Management; in: Journal of Lending and Credit Risk

Managment; February 1998.

DUFF95a Duffee, G. R.: On Measuring Credit Risks of Derivative Instruments; in: Working Paper,

Federal Reserve Board, February 1995.

DUFF95b Duffee, G. R.: On Variation of Default Risk with Treasury Yields; in: Working Paper, Federal

Reserve Board, January 1995.

FHSS94 Flesaker, Bjorn; Hughston, Lune; Schreiber, Laurence; Sprung, Lloyd; Taking all the Credit;

in: Risk, 7 (9) September 1994.

FRRM97 Fabozzi, Frank J.; Ramsey, Chuck; Ramirez, Frank R.; Marz, Michael: The Handbook of

Nonagency Mortgage-Backed Securities; New Hope PA 1997.

FRSK97 Fooladi, Iraj J.; Roberts, Gordon S.; Shinner Frank: Duration for Bonds with Default Risk;

in: Journal of Banking & Finance, 21, 1997.

GAST92 Gastineau, Gary L.: Dictionary of Financial Risk Management; Chicago, Cambridge 1992.

GEJO84 Geske, Robert; Johnson, H. E.: The Valuation of Corporate Liabilities as Compound Options:

A Correction; in: Journal of Financial and Quantitative Analysis, 19, 1984.

GEKR Gerdsmeier, Stefan; Krob Bernhard: Kundenindividuelle Bewertung des Ausfallrisikos mit

dem Optionspreismodell; in: Strategisches Bankcontrolling —Eine neue Herausforderung;

Publikation der DG Bank, Frankfurt am Main.

GESK74 Geske, Robert: The Valuation of Corporate Liabilities as Compound Options; in: Journal of

Financial and Quantitative Analysis, 19, 1984.

GLAN94 Glantz, Morton: Loan Risk Management; Chicago, Cambridge 1994.

GREN96 Grenadier, Steven R.: Leasing and Credit Risk; in: Journal of Financial Economics, 42 (3);

November 1996.

HALB94 Halbherr, Philipp: Neuere theoretische Grundlagen zur Fundierung eines modernen

Kreditmanagements; in: Informations- und Arbeitstagung des Instituts f ¨ ur schweizerisches

Bankwesen der Universit ¨ at Z¨ urich; Z¨ urich 1994.

HENN94 Henninger, Ute: Branchenrisikoeinsch¨ atzung im Rahmen der Bonit ¨ atsbeurteilung von

kommerziellen Kreditkunden; Bern, Stuttgart, Wien 1994.

HOF95 Hof, Michael: Kreditderivate —Aktive Steuerung von Kreditrisiken; Vortrag am 14.09.1995

im Rahmen der IIR-Fachkonferenz “Kreditmanagement”; Publikation der DG Bank, Frank-

furt am Main.

HOSI84 Ho, Thomas S. Y.; Singer, Ronald F.: The Value of Corporate Debt with a Sinking-Fund

Provision; in: Journal of Business, 57; 1984.

HULL97 Hull, John C.: Options, Futures, and Other Derivatives, third edition; Upper Saddle River

1997.

H

¨

URL93 H¨ urlimann, Werner: Risiko —aus morphologischer Sicht; in: io Management Zeitschrift, 62;

Z¨ urich 1993.

HUWH92 Hull, J.; White A.: The Price of Default; in: Risk, September; 1992.

Bibliography 165

HUWH95 Hull, J.; White A.: The Impact of Default Risk on the Prices of Options and other Derivative

Securities; in: Journal of Banking and Finance, 19 (May); 1995.

HSIA81 Hsia, Chi-Cheng: Optiomal Debt of a Firm: An Option Pricing Approach; in: Journal of

Financial Research, 4; 1981.

JATU95 Jarrow, R. A.: Turnbull, S. M.: Pricing Options on Derivative Securities Subject to Credit

Risk; in: Journal of Finance, 50; 1995.

JOST87 Johnson, H.; Stulz, R.: The Pricing of Options under Default Risk; in: Journal of Finance,

42; 1987.

KAEL98 Kaelhofer, Stephen: Portfolio Management of Default Risk; in: http://www.kmv.com; 1998.

KARO93 Karolyi, Andrew G.: A Bayesian Approach to Modeling Stock Retrun Volatility for Option

Valuation; in: Journal of Financial and Quantitative Analysis, 28 (4), December 1993.

KILG94 Kilgus, Ernst. Strategisches Bank-Management, Bern, Stuttgart, Wien 1994.

KNAB95 Knabenhans, Walter: Die Transparenz im Markt muss erh¨ oht werden —

¨

Uber die Zukunft,

Chancen und Risiken des Derivativmarktes; in: Finanz und Wirtschaft Nr. 18, 04.03.1005;

Z¨ urich 1995.

KREY91 Kreyszig, Erwin: Statistische Methoden und ihre Anwendung, 7. Auﬂage; G¨ ottingen 1991.

KRRO92 Kremer, Joseph W.; Roenfeldt, Rodney L.: Warrant Pricing: Jump-Diffusion vs. Black–

Scholes; in: Journal of Financial and Quantitative Analysis, 28 (2), June 1992.

LEE81 Lee, C. Jerons: The Pricing of Corporate Debt: A Note; in: Journal of Finance, 36, 1981.

MANZ98 Manz, Felix: Prozessorientiertes Kreditmanagement; Bern, Stuttgart, Wien 1998.

MASU81 Mason, Scott P.; Sudipto, Bhattacharya: Risky Debt, Jump Processes, and Safety Covenants;

in: Journal of Financial Economics, 9, 1981.

MEIE96 Meier, Christian: Lehren aus Verlusten imKreditgesch¨ aft Schweiz; bern, Stuttgart, Wien 1996.

MERT73 Merton, R. C.: Theory of Rational Option Pricing; in: Bell Journal of Economics and

Management Science, 4, Fr¨ uhling 1973.

MERT74 Merton, R. C.: On the Pricing of Corporate Debt: the Risk Structure of Interest Rates; in:

Journal of Finance, 2, 1974.

MOOD90 Moody’s Investors Service: Einf ¨ uhrung Credit-Ratings amKapitalmarkt; Frankfurt, London,

New York, Paris, San Francisco, Sydney, Tokyo 1990.

M

¨

ULL90 M¨ uller, J¨ org: Branchenkennzahlen zur Bonit ¨ atsbeurteilng von Klein- und Mittelun-

ternehmungen —Eine empirische Studie anhand von Zahlenmaterial der SKA; bankinterne

Studie der SKA; Z¨ urich 1990.

NATE94 Natenberg, Sheldon: Option Volatility and Pricing; Chicago, Cambridge 1994.

NEWT93 Newton, Brian: Modeling Credit Risk; New York 1993.

NZZ96 – Acra statt Abracadabra, Das neue Risikosystem des Schweizerischen Bankvereins; Neue

Z¨ urcher Zeitung, Nr. 218, Seite 29; Z¨ urich 1996.

– Altlasten-Beseitigung bei der CS Holding; Neue Z¨ urcher Zeitung, Nr. 298, page 19; Z¨ urich

1996.

PASS94 Passardi, Adriano: Bank-Management und Bank-Kostenrechnung; Bern, Stuttgart, Wien

1994.

R

¨

ASS93 R¨ ass, Hugo E.: Die Resrukturierung von Unternehmen aus der Sicht der Kreditgebenden

Bank; Bern, Stuttgart, Wien 1993.

RITR98 Ritchken, Peter; Trevor, Rob: Pricing Options Under Generalized GARCH and

Stochastic Volatility Process; in: Journal of Finance, forthcoming, http://www.cob.ohio-

state.edu/htbin/htimage/˜ﬁn/journal/jf.conf?94,154.

RSSO93 Reswick, Bruce G.; Sheikh, Aamir M.; Song, Yo-Shin: Time Varying Volatilities and Calcu-

lation of the Weighted Implied Standard Deviation; in: Journal of Financial and Quantitative

Analysis, 28 (3); September 1993.

SAVE96 Savelberg, Albert H.: Risikomanagement mit Kreditderivaten; in: Die Bank, 6, 1996.

SCM

¨

U98 Scholl, Martin; M¨ uller, Roger: Risikogerechte Preise; in: Schweizer Bank; 1; 1998.

SCHW82 Schwartz, Eduardo S: The Pricing of Commodity-Linked Bonds; in: Journal of Finance, 37;

1982.

166 Risk-adjusted Lending Conditions

SDHJ97 Scott Docking, Diane; Hirschey, Mark; Jones Elaine: Information and Contragiven Effects

of Bank Loan–Loss Reserve Announcements; in: Journal of Financial Economics, 43; 1997.

SMIT76 Smith, Clifford W. Jr.: Option Pricing; in: Journal of Financial Economics, 3; 1976.

SWDI92 Swidler, Steve; Diltz, David J.: Implied Volatilities and Transaction Costs; in: Journal of

Financial and Quantitative Analysis, 27 (3); September 1992.

TRTU96 Tripi, R.; Turban, E.,: Neural Networks in Finance and Investing; in: http://www.sigma-

research.com/bookshelf/rtbookn2.thm.

VASI84 Vasicek; Oldrich A.: Credit Valuation; in: http://www.kmv.com; 1984.

VIZA96 Vintschger von, R.; Zaker, S.: Kreditrisiken sind gef¨ ahrlicher als Marktrisiken; in: Finanz

und Wirtschaft, 21.09.1996; Z¨ urich 1996.

VOLK93 Volkart, Rudolf: Beitr¨ age zur Theorie und Praxis des Finanzmanagements, 5. erweiterte

Auﬂage; Z¨ urich 1993.

VOLK98/1 Volkart, Rudolf: Optionswertbildung und Optionspreistheorie —Versuch einer integrierten

‘didaktischen’ Darstellung; in: Working Paper Serie (Institut f¨ ur schweizerisches Bankwesen

der Universit¨ at Z¨ urich), Z¨ urich 1998.

VOLK98/2 Volkart, Rudolf: Optionstheoretische

¨

Uberlegungen zur Kapitalstrukturgestaltung und zur

Kreditﬁnanzierung; in: Working Paper Serie (Institut f¨ ur schweizerisches Bankwesen der

Universit¨ at Z¨ urich), Z¨ urich 1998.

VOLK99 Volkart, Rudolf: Optionstheoretische Analyse von Fremdkapitalkontrakten und alternative

Credit Risk Spread-Bestimmungen —eine integriete Darstellung; in: Working Paper Serie

(Institut f¨ ur schweizerisches Bankwesen der Universit¨ at Z¨ urich), Z¨ urich 1999.

WONG97 Wong, Kit Pang: On the Determinants of Bank Interest Margins under Credit and Interest

Rate Risks; in: Journal of Banking and Finance, 21; 1997.

ZELL83 Zellweger, Bruno:

¨

Uberwachung kommerzieller Bankkredite; Bern, Stuttgart 1983.

ZIMM98 Zimmermann, Heinz: C = S ×N(d

1

) −X · e

−r.(T−t)

×N(d

2

); in: Schweizer Bank; 1998.

ZKB96 Z¨ urcher Kantonalbank: Immobilienmarkt Z¨ urich; Eigenverlag; Z¨ urich 1996.

Index

acceptability of debt servicing, 112–115

annual rate of interest, 66

annual repayment, regular, loans on, 36

annual volatility, 66–67

approximate solution, 48–49

reliability of, 49–50

assets, 54

return on, 108–110

volatility of, 121

assumptions, model, 11–12

balance sheets, analysis, 123

bankruptcy

expectation value of loss in, 20–21

privileged salary claims and, 80–82

risk of, 78–79

bankruptcy-resistant debt rate, 115

Basel Committee for bank supervision, 12, 13,

145

behavioural risks, 3, 9

Black/Scholes approach cf KMV approach, 59–60

bonds, 80

breakdown distribution probability value, 80

breakdown distribution rate probability, 78

Bretton Wood agreements, 15

call put parity, 61

Cantonal Bank of Z¨ urich, 6, 135, 143, 151

capital asset pricing model (CAPM), 15, 68, 71,

121, 122, 123

cash ﬂow statements, analysis, 122–123

chaos theory, 15

clothing business, turn-around situation, 127–129

company’s value, 62

conﬂict between bank and borrowers, 125–126

continuous-time ﬁnance, 15–16

continuous-time framework, 15

corrected credit shortfall risk, 81–82, 85

correction factor, 38, 45, 69, 70

country risk, 14

covered loans, shortfall risk of, 91–101

borrower vs collateral correlation, 93–97, 143

derivation of correlation, 93–94

maximum value, 95–97

minimum value, 94–95

example, 99–101

option theory approach, 91–93

vs uncovered loans, 92, 98

credit balance assets, 54

credit limits, exceeding, 40

credit risk, 14, 16, 124

credit shortfall risk, 64, 66

inﬂuence of individual parameters, 71–77

Credit Suisse Group, 6, 7

credit-worthiness, 5

credit-worthiness key ﬁgure, deﬁning, 21

credit-worthiness risk, 9

current account credit, 26–40

current account credit write-off risk hedging rate,

37

current account loans, 52

customer considerations, 145

data limitations, 12

death, risk of, 54, 71

debt rate, 65, 67–68, 121

discount rate, 71, 123–124

discounted free cash ﬂows, 121

discretionary income, 71

earnings, declining, companies with, 114

earnings statements, analysis, 122

168 Index

effective loan interest rate, 45

effective proﬁt contribution rate, 43, 45

effective shortfall risk hedging rate, 43–44, 45

equity, return on, 108–10

exchange rate risk, 14

ﬁnancing companies, 142–143

ﬁnancing cost rate, 8

ﬁnancing rate, 36

ﬁxed advance

with complete repayment, 50–51

with partial repayments, 51

with regular repayments, 36

without repayments, 33–35

ﬁxed interest loan without repayments, 47–48

approximate solution, 48–50

free cash ﬂow, 68

gamma function values, 69, 70

insurance premiums, 139, 140

interest rates

acceptability of, 112

risk, 14

invalidity, risk of, 54

inventory limits, 39

investment income, 54

IT support, 143

jump-diffusion model, 83

KMV Corporation approach, 59–60

legal interest rate cap, 71

liquidation value, 121

liquidity risks, 3

loan assessment, 40–41, 79–80

loan assessment ofﬁces, 141–142

loan business, problems in, 7–8

loan combinations, 103–117

partially covered loans, 108

several loans, general case, 107–108

three loans, 107

two loans, 105–106

loan commissions, 8

loan exposure models, 13–17

classical methods, 13, 14

modern credit risk analysis, 13, 14–16

loan granted indeﬁnitely, 29

loan interest rate, deﬁnition, 8

loan interest rate model, 8–9

loan market value, 33

loan supervision, 114–115

loan terms, determination, 124

losses on loans, 5, 6

machine tools company, 129–130

management agreement, 54

marginal interest rate, 103–105, 115

market price of risk, 15

Market Risk Amendment, 12

market risk premia, 15

market risks, 3, 14, 16

market value, 60–61

maturity transformation, 28–29

maximum debt, 114, 115

maximum return on equity, 108–112

maximum shortfall risk covered, 44

mean shortfall risk, 48

minimum loan interest rate, 42, 45

Monte Carlo simulations, 14

mortgage, 143

business 1985–99, case study, 135–138

combination of loans, 116–117

shortfall risk, 99–100

net present value of loan transaction, 11

neural networks, 14

operational risks, 3, 9, 14

option theory approach, 4

basic formulae, 60–63

limits to application, 82–84

organisational measures, 143–144

partially covered loans, 108

personnel, specialist, 141–142

portfolio of securities, loans against, 143

private debtors, 71

privileged salary claims in case of bankruptcy,

80–82

probability, 31

probability model, 27–31

probability of cash ﬂows being fulﬁlled, 27–28

proﬁt contribution rate, 9, 11, 36–37

property assets, 54

property price index, 135–136, 151

rating system

ampliﬁcation, for very competitive markets, 23

need for, 19

in terms of ﬁgures, example, 21–23

repayment, acceptability of, 113

return on assets, 108–110

Index 169

return on equity, 108–110

return rate, 108–110

risk-adjusted loan interest rate per annum, 84

risk-adjusted values, derivation of, 63–67

risk assessment for lending to a company

analysis of cash ﬂow statements, 122–123

analysis of balance sheets, 123

analysis of earnings statements, 122

conﬂict between bank and borrowers, 125–126

determination of credit risk, 124

determination of discount rate, 123–124

determination of relevant loan terms, 124

determination of volatility, 124

overall view of procedure, 121–122

prudence in case of new loans/borrowers,

125

risk insurance premium, agreed, 140

risk measurement, 123–124

risk premium rate, 38–39

risk surcharge, model for calculating, 9–11

salaried employment, income from, 54

salary claims in case of bankruptcy, 80–82

sales staff, 142

self-employment, income/unearned income

from, 54

sensitivity testing, 13

ship mortgages, 130–135

banks’ loan decision, 130–131

comparison of bank/model assessment, 134

limitation of loan risk, 134–135

loan risk assessment by banks, 131

loan risk determination according to model,

131–134

model information, 131

starting position, 130

shortfall risk, 3, 31

deﬁning in terms of ﬁgures, 20–21

deﬁnition, 4

shortfall risk hedging rate, 9, 33–46

in general case of variable shortfall risk,

47–52

Compiled by Annette Musker

standard rate of interest, 10–11

stochastic calculus, 15

stockmarket risk, 14

success chance, 29–31

successor company provision, 129–130

survival chance, 31

Swiss Bank Corporation, 6

term of loan, 121

testing, model, 12–13

tests of hypotheses, 139–141

three loans, 107

transition matrices, 14

turn-around situation, 127–129

two loans, 105–106

uncovered loans, shortfall risk on, 53–55, 59–89

companies, 55

company with continuous business

development, 85–87

company with a poor ﬁnancial year, 87–89

covered/uncovered loans to same borrower, 98

inﬂuence of individual parameters on, 71–77

private clients, 53–54

vs covered loans, 92, 98

unemployment beneﬁt, rate of, 71

unemployment risk, 54, 71

validation, model, 12

value

of a company, 67–69, 70

of the debts, 62

variable shortfall risk, 47–52

volatility, 69–70

annual, 66–67

of assets, 121

determination, 124

whole option theory, 70

winding up ofﬁces, 142

zero bond approach, 66

**Risk-adjusted Lending Conditions
**

An Option Pricing Approach

Werner Rosenberger

**70+ DVD’s FOR SALE & EXCHANGE
**

www.traders-software.com www.forex-warez.com www.trading-software-collection.com www.tradestation-download-free.com

Contacts andreybbrv@gmail.com andreybbrv@yandex.ru Skype: andreybbrv

.

Risk-adjusted Lending Conditions .

2: New Markets and Products Carol Alexander (ed) Interest-Rate Option Models: Understanding.Wiley Finance Series Measuring Market Risk Kevin Dowd An Introduction to Market Risk Measurement Kevin Dowd Behavioual Finance James Montier Asset Management: Equities Demystiﬁed Shanta Acharya An Introduction to Capital Markets: Products. ELKS. DECS and Other Mandatory Convertible Notes Izzy Nelken (ed) Options on Foreign Exchange. Revised Edition David F DeRosa Volatility and Correlation in the Pricing of Equity. Lyons. Price and Manage Credit Risk Didier Cossin and Hugues Pirotte Dictionary of Financial Engineering John F. Measuring and Hedging Operational Risk Marcelo Cruz Monte Carlo Methods in Finance Peter J¨ ckel a Building and Using Dynamic Interest Rate Models Ken Kortanek and Vladimir Medvedev Structured Equity Derivatives: The Deﬁnitive Guide to Exotic Options and Structured Notes Harry Kat Advanced Modelling in Finance Using Excel and VBA Mary Jackson and Mike Staunton Operational Risk: Measurement and Modelling Jack King Advance Credit Risk Analysis: Financial Approaches and Mathematical Models to Assess. 1: Measuring and Modelling Financial Risk Carol Alexander (ed) Risk Management and Analysis vol. Marshall Pricing Financial Derivatives: The Finite Difference Method Domingo A Tavella and Curt Randall Interest Rate Modelling Jessica James and Nick Webber Handbook of Hybrid Instruments: Convertible Bonds. Strategies Participants Andrew Chisholm Hedge Funds: Myths and Limits Francois-Serge Lhabitant The Manager’s Concise Guide to Risk Jihad S Nader Securities Operations: A Guide to Trade and Position Management Michael Simmons Modelling. Preferred Shares. FX and Interest-Rate Options Riccardo Rebonato Risk Management and Analysis vol. Analysing and Using Models for Exotic Interest-Rate Options (second edition) Riccardo Rebonato .

Risk-adjusted Lending Conditions An Option Pricing Approach Werner Rosenberger .

West Sussex PO19 8SQ. UK. Chichester. [Risikoad¨ quate Kreditkonditionen.Published 2003 by John Wiley & Sons Ltd. the services of a competent professional should be sought. or faxed to (+44) 1243 770620. England Telephone (+44) 1243 779777 Copyright C 2003 Werner Robert Rosenberger First edition published in German by Paul Haupt Verlag. 2 Clementi Loop #02-01. USA Jossey-Bass. — (Wiley ﬁnance series) Includes bibliographical references and index. Etobicoke. except under the terms of the Copyright.co. English] a Risk-adjusted lending conditions : an option pricing approach / Werner Rosenberger . I.uk or www. photocopying. Singapore 129809 John Wiley & Sons Canada Ltd. London W1P 0LP. CA 94103-1741. Requests to the Publisher should be addressed to the Permissions Department.com All Rights Reserved. Designs and Patents Act 1988 or under the terms of a licence issued by the Copyright Licensing Agency Ltd. Title. paper) 1. Canada M9W 1L1 Wiley also publishes its books in a variety of electronic formats.uk. II. scanning or otherwise. ISBN 0-470-84752-2 (alk. electronic. Jin Xing Distripark. The Atrium.co. If professional advice or other expert assistance is required. D-69469 Weinheim.co. Other Wiley Editorial Ofﬁces John Wiley & Sons Inc. USA Wiley-VCH Verlag GmbH. Some content that appears in print may not be available in electronic format.wiley. John Wiley & Sons Ltd.A3 R67 2002 332. 4 Vaughan Avenue. Australia John Wiley & Sons (Asia) Pte Ltd. Risk management — Mathematical models. Chichester. West Sunsex PO19 1UD. Germany John Wiley & Sons Australia Ltd. It is sold on the understanding that the Publisher is not engaged in rendering professional services. Hoboken.1 753 0681 — dc21 British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library ISBN 0-470-84752-2 Typeset in 10/12pt Times by TechBooks. stored in a retrieval system or transmitted in any form or by any means. 3. cm. Option (Finance) — Prices — Mathematical models.uk Visit our Home Page on www. [translated by Christopher Massy-Beresford]. Southern Gate. Credit — Management — Mathematical models. 22 Worcester Road.wiley. Ontario. Series. New Delhi. Pappelallee 3. No part of this publication may be reproduced. 33 Park Road. 2. without the permission in writing of the Publisher. Queensland 4064. Berne. p. 90 Tottenham Court Road. recording. 2002031125 . Werner.. Guildford and King’s Lynn This book is printed on acid-free paper responsibly manufactured from sustainable forestry in which at least two trees are planted for each one used for paper production. San Francisco. Bafﬁns Lane. HG6024. 989 Market Street. India Printed and bound in Great Britain by Biddles Ltd. This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. Switzerland in paperback in 2000 Translated by Christopher Massy-Beresford. NJ 07030. Milton. London W6 0XS Email (for orders and customer service enquiries): cs-books@wiley. mechanical. 111 River Street. England or emailed to permreq@wiley. Library of Congress Cataloging-in-Publication Data Rosenberger.

For my family .

.

2 Narrowing the subject down.2 Deﬁning shortfall risk in terms of ﬁgures 2.5 Loan interest rate model 1.3 The model presented here seen in relation to previous models 2 Rating system 2.9.1 The problem 1.7 Assumptions 1.3 The insurance concept 1.6 Model for calculating risk surcharge 1.5 Ampliﬁcation of the rating system for very competitive markets xiii xv 1 3 3 3 5 7 8 9 11 12 13 14 14 16 19 19 20 21 21 23 .9 Loan exposure models 1.2 Modern credit risk analysis based on ﬁnancial theory 1.Contents Preface 1 Preface 2 Part I Outline 1 Introduction 1.3 Deﬁning the credit-worthiness key ﬁgure 2.4 Example of a rating system in terms of ﬁgures 2.1 The need for a rating system 2.9.8 Testing the model 1.9.4 Types of problem in the context of loan business 1. setting the objective and subdividing it 1.1 Classical methods 1.

7 Results and conclusions 6 Shortfall risk on uncovered loans on the basis of statistics 6.3.6.1 Fixed interest loan without repayments 5.2 Income from salaried employment 6.6.2 Companies 25 27 27 28 29 29 29 31 33 33 36 36 36 40 41 42 43 43 44 45 45 47 47 48 49 50 51 52 52 53 53 53 54 54 55 .2 Effective proﬁt contribution rate 4.1.3 Loans on regular annual repayment 4.1 Unearned income and income from self-employment 6.1.1.6.7 Results and conclusions 4.2 Fixed advance with regular repayments 4.6 Current account loans 5.4 Results and conclusions 4 Calculation of the shortfall risk hedging rate in the special case of shortfall risks being constant 4.4 Fixed advance with complete repayment 5.5 Loan assessment 4.3 Conclusions 3.1 The case of a loan being granted indeﬁnitely 3.2 Maturity transformation 3.1 Fixed advance without repayments 4.1 Minimum loan interest rate 4.3 Effective shortfall risk hedging rate 4.1 Determining the probability of cash ﬂows being fulﬁlled 3.2 Approximate solution for ﬁxed interest loan without repayments 5.8 Example 5 Calculation of the shortfall risk hedging rate in the general case of variable shortfall risk 5.5 Fixed advance with partial repayments 5.4 Current account credit 4.6 Conclusions 4.3.3 Reliability of the approximate solution 5.1 Private clients 6.6.2 Reﬂections on the success chance ε(n) 3.viii Contents Part II Mathematical Foundations of the Model 3 Probability model: Development of ψ j 3.3 Investment income and assets 6.4 Maximum shortfall risk covered 4.

2 Correlation between the shortfall risk of the borrower and the shortfall risk of the collateral 8.5 Results and conclusions 8.2 Two loans 9.2.2 Volatility 7.3 Derivation of risk-adjusted values 7.4 The general case of several loans 9.12.1 Shortfall risk of a covered loan on the basis of the option-theory approach 8.3 Three loans 9.4.11 Results and conclusions 7.5 Partially covered loans 57 59 59 60 63 67 67 69 71 71 78 79 80 80 82 84 85 85 87 91 91 93 93 94 97 98 99 99 103 103 105 107 107 108 .2.10 Limits to the application of the option theory approach 7.6 Example 9 Calculation of the combination of loans with the lowest interest costs 9.Contents ix Part III Option-Theory Loan Risk Model 7 Shortfall risk on uncovered loans to companies on the basis of an option-theory approach 7.6 Risk of bankruptcy and breakdown distribution 7.4 Covered and uncovered loans to the same borrower 8. together with further elaboration 7.7 Loan assessment 7.5 Inﬂuence of individual parameters on the credit shortfall risk 7.3 Shortfall risk of the covered loan 8.4 Deﬁnition of the values for the solution formula 7.3 Private debtors 7.1 Difference in approach between Black/Scholes and KMV.4.12.12 Examples 7.1 The value of the company and its debt rate 7.2 Example of a company with a poor ﬁnancial year 8 Loans covered against shortfall risk 8.2 Derivation of basic formulae 7.2 Value area of the efﬁciency of the correlation 8.1 Marginal interest rate 9.9 Consideration of privileged salary claims in the event of bankruptcy 7.1 Example of a company with continuous business development 7.4.1 Derivation of the correlation 8.8 Bonds 7.

6 Limitation of loan risk 11.8 9.3 Assessment of the loan risk by the banks 11.3.9 Maximum return on equity Acceptability of debt servicing 9.8 Determination of credit risk 10.3.2 Implementation in practice 12.1 Overall view of the procedure 10.2 Statistical bases 121 121 122 122 123 123 124 124 124 125 125 127 127 129 130 130 130 131 131 134 134 135 139 139 141 141 142 .4 Consequences for companies with declining earning 9.7.10 Possible causes of conﬂict between bank and borrower when the model is applied 11 Applications 11.4 Analysis of balance sheets 10.3 Analysis of cash ﬂow statements 10.5 Consequences for loan supervision Results and conclusions Example 108 112 112 113 114 114 114 115 116 119 Part IV Implementation in practice 10 Procedure — according to the model — for assessing the risk in lending to a company 10.3 Ship mortgages: Risk limitation 11.3.5 Determination of the discount rate 10.x Contents 9.3.1 Specialist personnel 12.3 Maximum debt 9.1 Tests of hypotheses 12.2 Acceptability of repayment 9.3.1 Starting position 11.9 Prudence in the case of new loans/borrowers 10.7 Determination of volatility 10.6 Determination of relevant loan terms 10.1 Specialist clothing business: Turn-around situation 11.7 9.4 Determination of loan risk according to the model 11.7.2 Analysis of earnings statements 10.7.4 Mortgage business 1985–99 12 Final considerations 12.2.2 The banks’ loan decision 11.7.5 Comparison of assessment between the bank and the model 11.7.2 Company trading in machine tools: Provision for successor company 11.1 Acceptability of interest rates 9.3.2.6 9.

4 Organisational measures Applicability of the method presented Customer considerations Open questions Closing remarks 143 143 144 145 145 146 147 149 151 153 155 159 163 167 Appendix 1: Notation Appendix 2: Excel worksheet Appendix 3: Property price index Appendix 4: Chapter 3 — Derivations Appendix 5: Chapter 4 — Derivations Appendix 6: Chapter 5 — Derivations Bibliography Index .2.3 12.3 IT support 12.5 12.2.4 12.Contents xi 12.6 12.

.

encouraged me to develop my assignment. I had to reduce the length of the ﬁrst plan considerably. in order to be able to keep the scope of this dissertation manageable from the view of the reader’s time as well as my own. .Preface 1 It is unusual to tackle a dissertation project 14 years after completing one’s studies. They u u had declared their willingness to submit my manuscript to critical examination. with whom I was able to have interesting discussions on the subject and who helped in working up the examples. who was its Director at the time. Professor Kilgus said on the basis of our discussion that he was prepared to act as my supervisor for this project. on the strength of how it has turned out. these have had to be left out — but they will form the subject of later publications. which compelled me to question things again and again. Although initial results have already been obtained on these subjects. for its indulgence in allowing me to attend some of the u normal lectures at the University of Z¨ rich during working hours. Included especially in this vote of thanks are many of my colleagues. I had already been a member of Credit Suisse’s management team for some years. I am extremely grateful to my supervisor for his thought-provoking comments. Thanks to his critical observations this book now contains many practical examples and advice to the reader. I can certainly put on record how pleased I am. At that point I had seriously underestimated the time that would be needed for this undertaking. It was originally envisaged that calculations of loan derivatives and assessment of a company’s equity would be examined using the methods described here. I thank my employer. and we agreed in the course of our conversations to bring this about in the form of a dissertation. Without him this book would have been substantially more theoretical and thus less easy to read and understand. however. to have brought this study into being. Credit Suisse. So that is the main reason why it has taken ﬁve years for this book to emerge. and was neither able nor willing to work there only part time. Professor Dr Ernst Kilgus. and provided me with many valuable ideas that have contributed to the success of this study. with the objective of making this work comprehensible and useful to a wider public. On the academic side I also owe many thanks to Professor Diethard Klatte and R¨ diger Frey u of the University of Z¨ rich and to Dr Philipp Halbherr of the Cantonal Bank of Z¨ rich. In my case it was an assignment at the Swiss Banking School that triggered it off. Looking back.

xiv Preface 1 When it came to this book. I managed to enlist the assistance of my brother. Wyssmann. Both have my sincere thanks for their invaluable help. P. My greatest thanks of all go to my partner Beatrice Wyssmann for the constant exchange of ideas. and on the other. and for her patience over its ﬁve-year gestation. which covered absolutely every aspect of the work. who scoured it with meticulous dependability for mistakes and inaccuracies. Rosenberger . I was able to count on the one hand on the considerable support of Mr H. u in its design and layout. May 2000 Walchwil Switzerland Werner R. J¨ rg Rosenberger. Thanks to her support I was able to bring it to fruition in a calm and orderly atmosphere. from the simplest to the most complicated.

Preface 2 Books about topics in modern ﬁnance theory have to be published in English today in order to reach an international audience. This is why I always wanted to publish my book in English too. The next step was to ﬁnd a translator. Her interest and her support have always been very encouraging and helpful. Samantha Whittaker. I thank Christopher for his excellent work. Stuart Benzie from McKinsey & Company. helped me to ﬁnd Christopher MassyBeresford for this job. Rosenberger . at a conference in Geneva in the autumn of 2000. Senior Publishing Editor at John Wiley & Sons. My special thanks go to my friend Carolina Schwyn-Villalaz. for all the support they provided me. I thank her and her successor. with whom I worked on a project in London during that time. September 2002 Walchwil Switzerland Werner R. She wanted to publish my book right from the beginning and was of great help to me at all times. I met Sally Smith.

.

Part I Outline Introduction Rating system .

.

S.2 NARROWING THE SUBJECT DOWN. 69] asserts that it is absolutely essential that these risks. among other things. which inevitably leave the bank’s shareholders with a poor impression. the more precisely can it — in the same way as insurance companies (cf. It is necessary to know the costs of each loan concerned precisely in order to be able to charge a price that fully covers its costs. The bank’s income depends on the prices for loans (see Section 1. to businesses and private individuals. can thereby be largely avoided (cf.5: proﬁt contribution rate). 1. that a high number of troubled loans brings about a signiﬁcant deterioration in cost efﬁciency. however. Unforeseen provisions and write-offs owing to surprise losses on lending business.5) that are charged to its borrowers. to assess these risks correctly for calculating the cost-covering price of any loan — but achieving this objective is far from simple (cf. including the costs that derive from the other risks listed above. both in the case of each individual loan and at the level of the bank’s operations as a whole. are identiﬁed. comprise: r shortfall risks r market risks r liquidity risks r behavioural risks r operational risks Kilgus [KILG94. These risks. The more precisely the bank knows its risks exposure. An important element in the costs of making loans arises from the risks that are inseparably bound up with such business. The greater the number of troubled loans that have to be managed. according to Kilgus [KILG94. assessed. The necessity.1 THE PROBLEM A bank can only pursue lending business proﬁtably if all the costs of providing loans can be covered by the income they produce. and will not cover lending to the public sector. as Berger and DeYoung [BEDE97] succeeded in showing. S. [DRZ198]). controlled and supervised in a comprehensive risk management plan. also results from the fact.3) — set up appropriate precautionary reserves.1 Introduction 1. [SDHJ97]. SETTING THE OBJECTIVE AND SUBDIVIDING IT This study will concentrate only on the domestic shortfall risks of any bank lending. It will be assumed that the costs of any loan. 66ff]. . It is therefore necessary to be able. Section 1. in its local currency. the higher the administrative costs owing to increased expenditure on supervision and winding up become (see also Section 1. are known.

Geske and others. in this case. these considerations will be further developed here. Here we go back to the original idea of Black and Scholes (cf. which demonstrates the properties speciﬁed. consistently. S.1). Undertaking such empirical testing as well would. As already mentioned. which a priori impair the accuracy of the model and make a rating system necessary. it is necessary to compare the model’s statements with real life on the basis of empirical testing. These will demonstrate that the Black/Scholes approach can be extended to reach a generalised solution that is applicable in real life. the probability of bankruptcy.69] by shortfall risk is meant.10 and 7. and it will be shown how the general case of non-constant shortfall risks can be linked up with the conclusions drawn in Chapter 4. This will take place in the context of the examples concerned. For this reason it has not hitherto been possible to evaluate enterprises with complicated structures on the liabilities side of the balance sheet. the bank’s demands. It is therefore emphasised quite clearly at this point that this study is concerned with developing a mathematical model. This assumption will be dropped in Chapter 5. Chapter 6 gives an indication of how the interrelations derived in the preceding expositions may be applied with the aid of facts determined statistically. The fundamental basis of the model is put forward in this ﬁrst chapter. we will fall back on the option-price theory approach of Black and Scholes [BLSC73. 637ff]. Thanks to the Nobel Prize won in 1997 by Black. as far as risk is concerned. This idea was taken up later by Merten. That shortfall risks are not in fact constant over time is demonstrated in Section 7. the fact that the borrower is either incapable of meeting. for example [GREN96] or [BRVA97]. [BLSC73]) of describing and evaluating debts and equity capital with the aid of their option-price theory. 4 and 5 — with the aid of probability calculus — the fundamental correlation between the risk-free rate of interest. or unwilling to meet. Figures 7. But all we will be doing here is mentioning.11). Scholes and Merten this approach is currently undergoing a minor renaissance [ZIMM98].4 Risk-adjusted Lending Conditions The objective is to develop a model that permits calculation of the lending costs which are caused by the shortfall risk. to complete the picture. however. Our expositions will thus link directly with their explanations (see Section 7.5 (cf. and the probability of shortfall on loan. the breakdown distribution probability value. in Chapter 4. The KMV Corporation in San Francisco. To attain this objective. the risk-adjusted rate of interest. Chapter 7 contains the crucial part of this study. As with all models that are intended to be of relevance in practice. will all be derived. the application of statistics as one possible way of solving the problem. by far exceed the scope of this study. We chose this course of action in order to show clearly how the conclusions were drawn. The main problem here was that the calculation of the rate of interest consistent with risk was always dependent on the risk-free rate of interest that was consistent with the time-scale. In the course of this we shall proceed. Following Kilgus [KILG94. on the unrealistic assumption that the shortfall risk is constant over the lifetime of any loan. The second chapter will tackle the problem of inaccuracies in the basic data involved. In Chapters 3. but never led to completely satisfactory results. Related approaches are also being applied to this (cf. All that can be done here is to make a qualitative comparison between the results and conclusions of the model. California (see [VAS184] and [KAEL98]) is adopting a similar approach. Under this the volatility of a company’s value is inferred via the . and experience in the lending business generally. S.

Their model does not. on the debt rate in relation to the market value. loans to private individuals that are consistent with risk may also be calculated. In the course of this study. With the aid of analogous conclusions. Loans to companies not listed on stock exchanges may also be assessed using it. the appropriate loan interest rate consistent with risk can be calculated individually for each part of the liabilities. by way of contrast. likewise proceeding from a stochastic process. Chapter 9 will demonstrate how loans with varied collateral may be combined to bring about debt ﬁnancing that is optimally structured in relation to risk for any company. This is always akin to walking a tight-rope. Chapter 8 will examine how a loan may be calculated on the basis of collateral and how the combination of borrower and collateral that is consistent with risk may be calculated correctly. To clarify the expositions so far. and if necessary to call the loan in. with losses on loans being regarded as individual cases. The expositions in Chapter 7 will show that a small detour leads to the objective. The application of the methods described will be elaborated in Chapter 11. the volatility of capitalised free cash ﬂows is used to assess the volatility of the company’s value using a model similar to the Black/Scholes equation. S. Which risks that result from the business of commercial lending represent the greatest and most commonly occurring ones thus becomes a matter of special importance to any banking institution. be suitably replaced by the credit shortfall risk. Chapter 12 will demonstrate to what extent the considerations applying to checking loans ought to be adjusted to line up with current normal procedures. As a result. how the empirical test mentioned of the methods presented here may be carried out. occurring unsystematically. however.Introduction 5 volatility of the stock exchange prices of listed companies. Traditionally it was the loan ofﬁcial’s job to identify problems of customer credit-worthiness in good time. that should be avoided as best as possible. It has therefore to be the aim of every bank.3 THE INSURANCE CONCEPT Hitherto banks have endeavoured to keep shortfall risks in their lending business as low as possible. among other things — according to Zellweger [ZELL83. The riskfree rate of interest and the rate of interest consistent with risk in the Black/Scholes equation may. with examples of loan transactions from actual banking practice. which has losses of proﬁts on loans and angry customers . Here the liabilities side can be as complicated as you like. the credit shortfall risk may be calculated depending on the volatility of the company’s market value. S. 3] also makes it the objective of his work to make a contribution to reducing and limiting shortfall risks. Chapter 10 will describe — in the manner of a cookery book — how to proceed in assessing loan risk according to this model.. for any given company. but pursues a similar solution approach. This calculation may be undertaken separately for any part of the liabilities. On the basis of the various credit shortfall risks. and what preconditions any bank would have to fulﬁl for their introduction. 4 and 5. and on the relevant term of any part of the total liabilities. use the Black/Scholes equation. i. 1. 1] — to keep the risks arising in connection with the exercise of individual banking transactions as small as possible.e. Meier [MEIE96. with the assistance of the conclusions from Chapters 3. with premature calling in on one side.

in that it establishes that ﬁnance companies there may be rather more inclined to take on higher risks. when granting loans. A study from the USA [CPSH98] comes to similar conclusions. involving the development not of methods of limitation. Here it is perfectly legitimate. as they are in the insurance business. S. We shall therefore adopt a fundamentally different approach here. as long as the overall relationship with the client is proﬁtable. In contrast to the USA there are. seen as failure because of the loss involved. But even in the case of calling in a loan in good time there is always uncertainty as to whether it was really justiﬁed in the circumstances. than banks. Two major Swiss banks in the lending business have already completed the move over to the insurance concept. It is in the very nature of things that any calling in of a loan that is too late is. 19] u the then Swiss Bank Corporation and the Credit Suisse Group have changed their working methods for accruing liability reserves profoundly. Losses on loans will not be considered as individual instances to be avoided. As soon as losses on loans become calculable. This does. The new task is to charge the right shortfall risk premium for each and every loan. and that it is in a position to prepare the information that will be needed. 29/NZZ96. It is at any rate important here that the running of higher risks must be associated with higher earnings. S. 218. The task of the bank’s loan assessment department therefore becomes fundamentally different. As could be seen in the Neue Z¨ rcher Zeitung [NZZ96. This is extremely unsatisfactory. Nr. however. but as costs that are. The client decides whether or not it is prepared to pay the shortfall risk premium included in the price for the loan. presuppose that skills in the bank’s accounting function have been appropriately developed. According to one study [DICH98] this does not appear . but of methods of calculating shortfall risks. with loan risks no longer representing extraordinary occurrences. Traditionally it had to decide whether or not the borrower’s ﬁnancial standing was up to the granting of credit at prices that were more or less set in advance. Now voices can be heard criticising this circumspection and reproaching the banks for exercising excessive prudence. in every case. The result of this is that the department of the bank responsible for sales will only close deals in which the borrower concerned is prepared to pay the price demanded. It is then up to the borrower to decide whether or not it is willing to pay the price demanded. to grant price concessions on loans. 298. Nr. u Many banks today have become substantially more circumspect in their lending than they were a few years back. it will be possible for banks to take on higher risks again ‘safely’.6 Risk-adjusted Lending Conditions as consequences. calculable. but now being treated as calculable costs. and on the other side calling-in operations that are too late and are associated with the resulting losses in receivables. if applicable in view of the borrower’s other banking business. This way the decision on whether any loan materialises no longer lies with the loan assessment department of the bank. no ﬁnance companies in Switzerland that would grant loans to companies to any extent worth mentioning. in much the same way as premiums are calculated for insurance policies. According ¨ to an article in the Schweizer Bank [SCMU98] it may also be assumed that the same is now true for the Cantonal Bank of Z¨ rich. As soon as credit risk costs can be calculated exactly. but with the client. it becomes possible to charge an appropriate risk-adjusted price for each loan. whichever way you look at it. however. It goes without saying that this course of action demands high standards of any bank’s sales departments. because of high losses on loans in the ﬁrst half of the 1990s.

Handling of delinquent loan business: this task is frequently undertaken by the loan assessment departments of a bank. including r r r r the costs of liquidity and market risks (cf. The departments of a bank that are responsible for its customers normally undertake this task. to provide all this information: in other words there is. however. Calculation of the running costs for the processing of a loan. in the lending business. 66ff]: a bank’s accounting department usually undertakes this task. Losses calculated in advance according to the . Negotiation of price with the borrower. It is.Introduction 7 to be the case in the USA. between the following types of problem. Calculation of the own resources requirement and/or additional liability reserves to cover statistical ﬂuctuations as actual losses appear in any given total lendings situation. including the costs of behavioural and operational risks (cf. At the Overall Banking Level r Preparation of a management information system for assessing total lendings. 66ff]): a bank’s treasury department usually undertakes this task. there are specialist units within the organisation responsible for this. The tasks at the overall banking level should be assigned to an ofﬁce that is responsible for the overall supervision and control of total lendings. It is not absolutely necessary for this that the individual types of problem be dealt with separately by different departments within the bank. This is essentially a question of the calculation of cluster risks and of the effects of diversiﬁcation. In the case of the Credit Suisse Group. A comparable investigation of the situation in Switzerland has not been undertaken.4 TYPES OF PROBLEM IN THE CONTEXT OF LOAN BUSINESS Our arguments so far rest on the assumption that one has to distinguish. and for evalr uating the work of the departments of the bank involved in the lending business described above. for example. S. interesting to note that this is the case among the banks in the Credit Suisse Group. [KILG94. no fundamental difference between management information systems for an insurance company’s motor liability insurance and a bank’s mortgage business. Banks should be led by the insurance concept in the building up of these organisational structures as well. as is normal in the case of any insurance company. Calculation of the shortfall risk costs of a loan: this is a question of a new task for the loan assessment departments of banks. These types of problem are: At the Individual Transaction Level r Calculation of the costs of reﬁnancing and underpinning a loan from own resources. taking the overall proﬁtability of the customer into account. for instance. [KILG94. Any management information system should thus be in the same situation. 1. S.

8 Risk-adjusted Lending Conditions model and losses actually incurred must be compared with each other on the basis of various parameters. . loans to debtors of good ﬁnancial standing may be underpinned with less of the bank’s own resources than loans to debtors of worse ﬁnancial standing.2. of statutory underpinning by own resources. from a business management point of view. however. irrespective of the debtor’s ﬁnancial standing. Thus a loan interest rate has to be calculated in the same way as the price for the loan. 1 In Switzerland today uncovered loans to companies have to be underpinned by own resources to an average extent of 8%. the price of which is expressed in the form of regular monthly payments. and the parameters further scrutinised for their relevance. and which is indeed paid by the borrower. as the division between the loan interest itself and the bank’s commission may in principle be undertaken at will. given level. does not mean that 8% of each individual loan has to be underpinned. for example. These costs accumulate in proportion to the amount of loan that is taken up. and the costs of market risks and liquidity risks. Swiss practice also involves charging ‘loan commissions’ to the borrower in the case of loans on current account. as long as on average the underpinning is still at least 8%. as already implicitly indicated in Section 1.1 Here it has been assumed that the bank’s treasury can calculate this rate on each loan at a ﬁxed. in order to complete the picture. however. 1. as is often to be observed in banking practice. leasing business and consumer credit. It would go beyond the remit of this present study to go more closely into the organisation of a management information system and into the calculation of the own resources that would be needed. At this juncture the intention is simply to point out the need concerned. The ‘loan commissions’ form an element of i. likewise determined from imputed interest rates. This simpliﬁcation may therefore be undertaken for the purposes of the model. also [SCMU98]): i = f + p+r i f p r = = = = loan interest rate ﬁnancing cost rate proﬁt contribution rate shortfall risk hedging rate (1. Exceptions are.1) Loan Interest Rate The loan interest rate is deﬁned here as the rate that is charged to the borrower as the price for the loan. Financing Cost Rate The ﬁnancing cost rate is deﬁned here as that rate. in this model. These regular monthly payments are. Our expositions on this are based on the following loan interest rate ¨ model (cf. in calculations internal to banks.5 LOAN INTEREST RATE MODEL The price of a loan is normally expressed in the form of a loan interest rate. the revenue from which will cover the costs of reﬁnancing. An integrated loan interest rate is assumed here for the sake of simplicity. It is rather that. That.

In return the borrower undertakes to pay the bank loan interest and repayments on the dates agreed. which do not accumulate in proportion to the amount of credit that is taken up. It will be explained in Section 4.2 that negative proﬁt contributions and earnings may also result from any loan. Shortfall Risk Hedging Rate The shortfall risk hedging rate is deﬁned here as whatever cost rate covers the shortfall risk costs of the loan concerned. increased administrative costs therefore lead also to higher loan interest rates. This leads to the fact that behavioural risk and shortfall risk are almost identical due to absence of borrowing power. As a more recent study shows [WONG97]. In the case of advances made at increased credit-worthiness risk even higher proﬁt contribution rates should be estimated because of the increased administrative expenditure (cf. In this a multi-stage proﬁt contribution rate calculation may be brought to bear. together with the few instances of fraudulent raising of loans. in addition to increased shortfall risk costs. [BEDE97]).6 MODEL FOR CALCULATING RISK SURCHARGE In any loan transaction the bank undertakes to pay the loan amount over to the borrower at a deﬁned point in time. Behavioural risks thus mostly surface only when the borrower has no further borrowing power. also Section 4. the behavioural risks and the operational risks. after deduction of the above-mentioned costs. should therefore be ascertained by statistical methods and may consequently be portrayed as part of the proﬁt contribution rate. We will demonstrate there how the effective proﬁt contribution rate is calculated. after deduction of shortfall risk costs. At ﬁrst therefore there is a cash ﬂow from the bank to the . in that only those transactions are entered into in which the margin. That the behavioural risks are reﬂected in the proﬁt contribution rate is connected with the following assumptions: normally the borrower is only deﬁned as a behavioural risk if it comes under sustained ﬁnancial pressure as a result of debt servicing charges.6. in the earnings from the loan concerned. This puts a ceiling on any bank’s lending activities. and permits an appropriate proﬁt (cf.6) to be earned on the loan. And in the case of advances with higher credit-worthiness risk. We assume that this rate is dependent on the ﬁnancial standing of the borrower. the ﬁnancing cost rate and the shortfall risk. Minor deviations resulting from this assumption.Introduction 9 Proﬁt Contribution Rate The proﬁt contribution rate is deﬁned here as the rate that covers the costs of handling the loan. if the following variables are quantiﬁed — the loan interest rate negotiated with the customer. there is an optimum interest margin for any risk-averse bank in consideration of credit-worthiness and market risks. 1. We make the assumption here that the ofﬁce in the bank responsible for sales has to obtain the highest possible loan interest rate in negotiations with the borrower. and therefore the highest possible proﬁt contribution rate: the revenue from the ﬁnancing cost rate results. permits a sufﬁcient proﬁt contribution.

the price of the risk-encumbered loan L is expressed in the nominal values of the cash ﬂows C j . if this total corresponds to the loan paid over. the cash ﬂows nominally agreed in the loan contract do not ﬂow to the bank. As will be shown in Section 7. In this exposition the standard rate of interest will therefore take over the function of the risk-free rate of interest in ﬁnancial market theory. Complicated loan arrangements in respect of amounts paid out and amounts paid back may be broken down into their individual components as such. We will go into this more closely in Chapter 3. in terms of value ﬁgures. the model for calculating the risk surcharge will now run as follows: n L= j=1 ψj · Cj (1 + i s ) j (1. the sum of the present values of the breakdown values of the cash ﬂows under the loan contract have to be determined. respectively. is fundamental.2) L n ψj Cj is = = = = = paid out loan amount number of periods of loan maturity probability of cash ﬂow C j cash ﬂow face value after period j risk-free standard interest rate Here the consideration that it is not just the face values.10 Risk-adjusted Lending Conditions borrower. The variable ψ j acquires decisive signiﬁcance. As the correctly calculated breakdown values of the cash ﬂows.3. Continuing. although it may remain the same for the term. we assume that there is a uniform risk-free standard rate of interest i s as a reference rate for each loan transaction.3) The reason for the standard rate of interest i s is the following: at a purely theoretical level a borrower might think that it presents no shortfall risk at all for the bank when any loan is granted. As demonstrated below. under the theory explained here. is independent of the standard interest rate and of the risk-free interest rate. There is thus absolutely certainty that the debt servicing will be performed according to contract. providing the statutory . the statutory underpinning by own resources may be varied according to ﬁnancial standing. but only their probability values which lie. S. Expressed mathematically. they must be discounted by the corresponding risk-free standard interest rate. however. which is calculated as follows: is = f + p (1. As already mentioned footnote 1. 340ff]. and then one or several cash ﬂows from the borrower to the bank. Because of the shortfall risk. may be regarded as implicitly more secure payments. in order to be able to assess the loan that has been paid over. which are already an expression of the correctly calculated shortfall risk. without being identical to it. between zero and their face value. We assume here that the risk rate is correctly calculated. The bank only has to charge such a debtor the reﬁnancing costs and an appropriate proﬁt contribution. but the breakdown values of the cash ﬂows that are discounted. The standard rate of interest does not have to be identical for all borrowers. this action is permissible as a borrower’s shortfall risk. The rate of shortfall risk is precisely zero. Following Volkart’s expositions on the assessment of ﬁnance contracts [VOLK93.

and can therefore be imputed in an integrated ﬁnancing cost rate for the loan transaction concerned. S. 3.7 ASSUMPTIONS At this point the assumptions made in this study are listed again: 1. if applicable calculated at a number of different levels or stages. 1. Customers of better ﬁnancial standing usually give rise to less loan assessment expenditure and vice versa. the appropriate earnings for the bank are therefore built into the calculation. 7. 5. the highest possible proﬁt contribution rate. and be covered by. starting from the proﬁt contribution normally achieved. if the total of the expectation values of the cash ﬂows. 2. at which it is the expectation values of the cash ﬂows. 69] is calculated. The shortfall risk hedging rate of a loan will have been correctly pitched. The revenue from the loan transaction concerned will be calculated via a calculation of proﬁt contribution rate. As the risk-free standard rate of interest in equation (1. For each loan transaction there is an integrated standard rate of interest as a reference rate.3) already contains a proﬁt contribution rate for the bank extending the credit. The shortfall risk hedging rate is dependent on the borrower’s ﬁnancial standing. behavioural and operational risks). which justiﬁes making such a distinction overall. 6. The net present value of the loan transaction is therefore [BRMY96. The price of a loan transaction may be expressed in one integrated loan interest rate that incorporates all the elements of the price. not thereby implied that no other risks exist (for example. on average. not their nominal values. 35]: n NPVCredit = j=1 ψj · Cj (1 + i s ) j −L (1. The only assumption here is that these risks have already been taken into account (see below).4) the requirement for a loan transaction consistent with risk may be formulated as follows: a loan transaction is reckoned to be entirely consistent with risk if its net present value calculated at the standard rate of interest is equal to zero.Introduction 11 requirements are. the behavioural and operational risks do not arise in proportion to the extent to which the credit is taken up. Only the shortfall risk as deﬁned by Kilgus [KILG94. Assumption 1 is a voluntary restraint for the purposes of this study. S. and have to be imputed into the revenue from.2) may be considered to be the investment the bank has to make in the loan transaction concerned. The proﬁt contribution rate does not have to be identical either. The costs of processing a loan. S. 35] the amount to the right of the equals sign in equation (1.4) Using equation (1. It is. are precisely equal to the loan amount extended under the loan agreement. 13. met. Following Brealey and Myers [BRMY96. however. The bank’s ﬁnancing costs and the costs of a loan transaction in relation to market and liquidity risks accrue in proportion to the amount of credit that is taken up. and the credit paid out to the left of the equals sign in equation (1. 4. that are discounted. liquidity.2) may be considered to be the present value adjusted for risk in respect of ﬁnancial standing. discounted by the standard rate of interest. market. . S.

for example [BRMY96. 1. The longer holding period. and the predictive nature of a credit risk model does not derive from a statistical projection of future prices based on a comprehensive record of historical prices. Where market risk models typically employ a horizon of a few days. this assumption may not be completely sustainable. a quantitative validation standard similar to that in the Market Risk Amendment would require an impractical number of years of data. It will nevertheless be made frequently in comparable situations. [BRMY96. Hence in specifying model parameters. at the time the loan agreement is concluded or the loan is made available. as preconditions of the loan. and to their various time-scales. Assumption 5 is self-evident. and has come to the conclusion that there cannot yet be any question of applying such models in the sphere of supervision. in order for it to be possible to make reliable use of this model. Complicated term structures of standard interest and/or discount rates under equation (1.2. [BCBS99. According to how many lines of credit exist between the borrower and the bank. S. in order to simplify our considerations (cf. S.2) may also be converted to one integrated rate. In the last few years. Most credit instruments are not marked to market. Assumption 2 is valid for the duration of the loan. 2] . The scarcity of the data required to estimate credit risk models also stems from the infrequent nature of default events and the longer-term time horizons used in measuring credit risk. at least approximately. credit risk models generally rely on a time frame of one year or more. and forms the main subject of this study.12 Risk-adjusted Lending Conditions Assumption 2 is in line with normal banking practice up until now. 1] Model validation: the validation of credit risk models is fundamentally more difﬁcult than the back-testing of market risk models. Two problem areas in particular led to this conclusion. By the same token. Assumptions 3 and 4 are in line with the current practice of any bank in terms of business management. Assumption 7 is not made to the absolutely fullest extent for each loan transaction. presents problems to model-builders in assessing the accuracy of their models. coupled with the higher conﬁdence intervals used in credit risk models. spanning multiple credit cycles. 36]. The forecasts made using it have to be capable of being checked against examples of losses on loans that actually occur. The relative size of the banking book — and the potential repercussions on bank solvency if modelled credit risk estimates are inaccurate — underscore the need for a better understanding of a model’s sensitivity to structural assumptions and parameter estimates. 646–649]). Assumption 6 is the fundamental assumption of the model. [BCBS99. in the course of calculating the rate of yield to maturity (cf. some banks have been attempting to subdivide this principle by introducing fees for administering loans and investigating credit status. S. Assumption 2 nonetheless implies no qualiﬁcation to the generality that such fees have traditionally only been applied. Data limitations: banks and researchers alike report data limitations to be a key impediment to the design and implementation of credit risk models. in view of the reservations outlined in Section 1. We will not go into the related difﬁculties in application here. credit risk models require the use of simplifying assumptions and proxy data.8 TESTING THE MODEL A mathematical model is only as good as the extent to which it is able to reﬂect real life. S. however. and this study is no exception. The Basel Committee for bank supervision has tested existing models for measuring lending risks.

To do this we must ﬁrst brieﬂy review the methods that have been used to date. 6] The introduction and above all the regular examination of a mathematical model for measuring credit risks represents a major challenge for any bank. The Basel Committee made it its primary concern to examine whether there are already models suitable for the purposes of supervision as required by law. the reporting process and other traditional features of the credit culture — will also continue to play a key part in the evaluation of a bank’s risk management framework. This suggests that the area of validation will prove to be a key challenge for banking institutions in the foreseeable future.Introduction 13 In addition the Basel Committee set out what precautions should be borne in mind when testing models. the rigour of stress testing.9 LOAN EXPOSURE MODELS To conclude this introduction. be more of a second step. however. [BCBS99. if a bank is indeed prepared to create the necessary preconditions for doing so now. . At present. Finally. and is at present still closely linked to development of principles. at this juncture. Only then may the extent to which such a model might also be suitable for use for the purposes of supervision. 1. its credit risks with a mathematical model. few banks possess processes that both span the range of validation efforts listed and address all elements of model uncertainty. or analysing the results of model output given various economic scenarios. it is important to note that the internal environment in which a model operates — including the amount of management oversight. [BCBS99. This may. and in turn to forecast. or verifying that the ex-ante estimation of expected and unexpected losses is consistent with ex-post experience. At present. that methods for testing mathematical models measuring credit risks have still to be developed. Some bank has ﬁrst to succeed in furnishing proof that it is in a position reliably to measure. methods such as sensitivity testing are likely to play an important role in this process. be examined. the model developed in this book is set in the context of loan exposure models existing hitherto. 50] In Chapter 12 we will explain what precautions should be borne in mind in order to meet the problems outlined above when applying the model presented here. S. S. The components of model validation can be grouped into four broad categories: (a) back-testing. and (d) ensuring the existence of independent review and oversight of a model. there is no commonly accepted framework for periodically verifying the accuracy of credit risk models. the quality of internal controls. In this respect it is indeed timely to emphasise that the method introduced here may only be examined for its validity in the future. (b) stress testing. r Methods based on modern ﬁnancial theory that has been developed since the 1970s. going forward. (c) assessing the sensitivity of credit risk estimates to underlying parameters and assumptions. as required by law. In this respect it is timely to stress. The previous models for deﬁning the shortfall risk of a loan can be classiﬁed into two groups: r Classical methods prior to the development of ﬁnancial theory.

p. exposures linked to derivatives to rise with the volatility of the markets. Rating agencies are standard sources for default probabilities. Second. and operational risk. One can expect. credit risk. in relation to the earlier models. they present some difﬁculties. historical correlations are difﬁcult to obtain empirically.9. 9–13]) Modern credit risk analysis. is along the line of the continuous development of ﬁnancial research on the integration of uncertainty. For instance Cossin and Pirotte [COPI01] give a good overall view of credit risk models in existence. To understand why the latter has been such a preoccupation in modern ﬁnance. for example. the investor faces risks that are categorised as market risks.1 and 1. and we accordingly refer the reader to the bibliography. and the needs that have increased with respect to them. At this juncture it is not.9.9.2 Modern Credit Risk Analysis Based on Financial Theory (see [COPI01.2 are based substantially on their compilation.1 Classical Methods (see [COPI01. Market risks integrate interest rate risk. As stressed by [DUFF95a]. for example [TRTU96]). The basic principle of this type of approach is to estimate (often independently) the value of the contract at possible default times. exchange rate risk and stockmarket risk. possible to go all that much into detail. These probabilities (usually organised in so-called transition matrices) consist of the probabilities of downgrade and upgrade by rating category. they rarely take into account the correlations among probabilities of default and estimates of possible losses. But it is also at such a time that probabilities of default will arise. end users tend to develop Monte Carlo simulations without taking into account the uncertainties in the models used to generate the estimates. pp. Although these methods are widely used in banks.9. Nonetheless. These calculations are now frequently used by professionals. but ﬁnancial theory has now provided us with more powerful analytical tools. They are still a useful basis of information to start from. all these methods face the major difﬁculty of being strongly dependent on historical estimates of credit risk dynamics. on the other hand. let us introduce some chronology about market risks. Methodologies have evolved from the calculations of mortality rates to the calculation of rating category migration probabilities.14 Risk-adjusted Lending Conditions The previous courses of action — divided into the two groups mentioned — are presented in outline in the following two subsections. Some try to overcome this difﬁculty by using advanced analysis methods such as neural networks (see. The expositions that follow in 1. Interest in market risks began ﬁrst with the development of stock exchanges and banking systems in .9. 1. Broadly speaking. 1.3. in Subsection 1. These correlations certainly affect the results. see also for a critical approach [DUFF95a/b]). 91]) Most of the classical literature on credit risk tends to bear on traditional actuarial methods of credit risk (see [CAOU98] for a survey of these. The model described in this book is put into context. their development. Unfortunately. country risk. Modern appraisal of credit risk follows directly from the advances that have been made for the management of market risks. however. Techniques used to forecast default probabilities for individual ﬁrms are described in [ALTM77].

there are no discontinuities at all in the evolution of the stock price. Suppose that we want to draw the evolution of the stock price and the terminal values that it can take in one month. Before its . Moreover. In the latter case. The dynamics of the unexpected part of the uncertainty is not deterministically speciﬁed from the beginning. The evolution happened on two grounds: ﬁnancial theory. ‘Market’ because the risk comes from market variables and also because. while in continuous time there is an inﬁnity of time steps guiding the stock price to its terminal value. to relate ﬁnancial theory and the continuous-time approach introducing the well-known continuous-time ﬁnance. The continuous-time framework is very useful because it enables much more easily closedform solutions to speciﬁc ﬁnancial problems to be obtained. as is the case with chaos theory. in order to ﬁnd a unique market risk premium for each factor. From the late 1970s. Let us take an example. The contribution of Merton resides in his capacity. Assumption 2: The stochastic processes generating the state variables can be described by diffusion processes with continuous sample paths. the classical present value of coupon payments and the simple calculation of durations and convexities appeared to be insufﬁcient to monitor and manage those risks. On the practitioners’ side. while the discrete approach is still of great help to visualise the choices to make through time. On the mathematical side. mostly driven from economics theory at that point. The contribution of stochastic calculus is ﬁrstly its capacity to produce a deterministic solution out of an uncertainty that is modelled as a random process. we have to choose the number of time steps up to the maturity of one month. the general hypothesis being made is that only systematic risk (the undiversiﬁable one) is priced. On the ﬁnancial theory side. exchange rates were then allowed to ﬂoat causing volatility in interest rates. The interest of academics in developing new theories to modelise the uncertainty of marketrisk phenomena has led to a sophisticated set of ﬁnancial tools inspired from mathematics and physics. In continuous time. This produced contracts that not only were sensitive to changes in market factors but also showed discontinuities in them. and the inclusion of sophisticated mathematics. much of today’s inheritance comes from the early introduction of stochastic calculus (well known in physics for its application to problems such as health propagation) into modern ﬁnance. In discrete time. But these choices are made on speciﬁc dates while they are made continuously in continuous time. Therefore. many forms of contracts were proposed to the investors to mitigate the increasing volatility on the market. most of the research attempted to give a value to the market price of risk or market risk premia.Introduction 15 most of the developed countries. the process does not execute jumps to two adjacents values but rather changes in a very small period of time to very small different values. at that time. with special clauses allowing them to be optionally protected against changes in the term structure. two basic assumptions have to be made to justify the use of continuous-time approaches in the portfolio selection problems of modern ﬁnance: Assumption 1: Capital markets are open at all times meaning that agents can trade continuously. stochastic calculus allows the reﬁning of the time space into inﬁnitesimal pionts as a limit of the discrete-time approach. ﬁnancial theory. many economic studies were undertaken giving rise to what has been called since. With the end of the Bretton Wood agreements. As noted in Merton’s articles.

One of them which is very relevant is the fact that differences between European currencies are vanishing with the appearance and global use of the Euro in ﬁnancial markets. not only the borrower’s shortfall risk but also the credit shortfall risk. From 1996. the standard rate of interest without taking the credit shortfall risk into account. In 1973. in line with shortfall risk. The contribution of Merton is substantial and visionary. the credit shortfall risk. which has a price therefore at time 0. This algorithm results in one receiving. we cannot stand on traditional actuarial approaches for the credit risk part to price those bonds. for each debt position in the company balance sheet. The investor is now allowed to react continuously to changes in the environment rebalancing and hedging his positions through time. Another reason is the huge movements in credit standing characterising the end of 1998. ﬁnancial theory was limited to static theories. This is a substantial advance on the Black and Scholes account [BLSC73]. precisely about credit risk.16 Risk-adjusted Lending Conditions emergence. Therefore. Black and Scholes were able to price such complex products as standard call and put options. That an extension of their model is able to achieve such results does. having this possibility means that the investor has a non-executed option on future allocation. The combination of these two bases leads to an algorithm that allows the risk of each individual debt position in any company balance sheet to be assessed on the basis of ﬁnancial theory (or. on the basis of option price theory). more precisely. Continuous-time ﬁnance allowed the restatement of previous problems dynamically. falling economies’ effects propagate strongly and quickly to high-grade economies. The wave of developments for market risks has engendered a sudden awareness in other ﬁelds. and the risk-adjusted loan interest rate. Here. we observe a continuously growing number of defaults occuring. From these . giving rise in the 1980s and 1990s to a rush into the design of derivatives products of increasing complexity. 1. On the one hand it is based on the groundwork developed in Part II of this book on the connections between the shortfall risk of the borrower. credit risk then becomes the main determinant of the spread of a corporate bond yield over the risk-free rate. market efﬁciency demands that the potentiality of credit losses must be accurately estimated and priced. This ability should be taken into account along with transaction costs to show how strategies can be optimised from the beginning. This interest also ﬁlls a need. since a linkage is directly being made between the interest rate risk and the credit risk. On the other hand it is based on the original Black/Scholes model [BLSC73] for the evaluation of company debts.3 The Model Presented Here Seen in Relation to Previous Models The model presented here has two bases. Moreover. for several reasons.9. Now that market risks seems to be well encompassed. showing how theories such as the CAPM are inﬂuenced if the investor is now allowed to behave dynamically. however. All these tools enable us today to price securities subject to these risks and to design sophisticated contingent claims on the same securities. With the globalisation wave. research has turned to credit risk. These connections are derived with the aid of classical probability calculus. also demonstrate how revolutionary their reﬂections already were. the breakdown distribution rate. Investors are thus far aware that credit risk is a real problem and that it cannot be measured and monitored on a standalone basis. For European currencies’ denominated bonds.

in order to determine the credit shortfall rate more precisely in the case concerned. it also allows for combination with the results of classical approaches. with its associated advantages. . for each position. When linked in with essential professional experience in lending business operations. it is thus possible to calculate the best possible estimations of loan exposure for given loans (in so far as this is.9). In so far as it may be necessary to do so. The borrower’s shortfall rate determined according to the model may now be combined with the modiﬁed breakdown distribution rate. for example. In this way the experienced professional obtains an effective instrument for assessing and evaluating loans. ex ante. agree with previous experience in the case of a concrete loan transaction — or that it has still to be adjusted (cf. our model thus allows for situations to be examined using a combination of classical methods and methods based on ﬁnancial theory. for instance. Owing to its ﬂexibility. Using the standard rate of interest. The method put forward here is thus indeed based on ﬁnancial theory. however. Section 7. Let us assume now that the breakdown distribution rate calculated according to the model does not.. The great advantage of these detailed results lies in that the fact that they may now be combined with classical ways of looking at things.Introduction 17 one can calculate. the probability value — according to the model — of the breakdown distribution rate. possible at all). we arrive at the risk-adjusted loan interest rate. where this is appropriate and meaningful.

.

1.1 THE NEED FOR A RATING SYSTEM Kilgus himself emphasises the need to subdivide a bank’s borrowers into various risk categories (ratings) [KILG94. Anything else would be crystal-ball gazing. but the future cannot be considered simply by extrapolating the past. The use of the maximum value for a group is consistent with the principle of conservatism. S. We will explain below why this applies here too. The question of what might be the optimum estimated accuracy versus the estimated expenditure needed to obtain it can only be answered by staying permanently close to what is going on in the marketplace. The fundamental difﬁculty in deﬁning a borrower’s future shortfall risk consists in putting forward an ex-ante forecast on the basis of ex-post facts. What this entails precisely is explained in Section 2. A numerical rating system is introduced in Section 2. 2. The same imputed shortfall risk is therefore allocated to all borrowers with the same rating. The reﬁnement of the rating system chosen depends essentially on how much cost it is intended to incur on the estimated accuracy of future shortfall risks. are beyond the scope of the science of business management.3. 70]. also [CART98]).2. The market’s sensitivity to price has to be reﬂected in the rating system. Borrowers are subdivided into groups with comparable shortfall risks. moreover. and that any other action would simply mean faking apparent exactitudes. The individually calculated shortfall risk is not used for calculating the loan price.4 and elaborated in Section 2. There is simply no easy answer to this. as reproducible experiments. borrowers that may not yet have any kind of track record. This brief reﬂection indicates right away that a borrower’s future shortfall risk cannot be deﬁned as precisely as one would like. and that a soundly deﬁned estimated value will always be required to a greater or less degree. On the other side of the coin. Historical facts do provide useful indications for this. and the credit-worthiness key ﬁgure is deﬁned in Section 2. There are. This in turn depends on how competitive a bank seeks to be in the market for any particular loan product. and what has to be done to deﬁne a borrower’s rating mathematically.5.4. such as are possible in the natural sciences. . more expenditure on estimation means higher processing costs and with that of necessity a larger proﬁt contribution rate p in the calculation of the price of the loan product. the shortfall risk is deﬁned numerically in Section 2. but the maximum value according to the rating level. The introduction of a rating system offers a way out of this dilemma (cf.2 Rating System The necessity for a rating system is explained in Section 2. This action is based on the realisation outlined above that future values can only be estimated.

2 DEFINING SHORTFALL RISK IN TERMS OF FIGURES It is assumed here that a shortfall risk ρ in terms of ﬁgures can be assigned to each borrower. at any time within any future period of time.1) Bankruptcy cases are far from meaning that all the money lent has been lost to the bank. or not be in such a position: there is no third option. The following correlation applies. of meeting its commitments to the bank in full. within the same period of time.20 Risk-adjusted Lending Conditions 2.6). Multiplied by the probability of bankruptcy occurring ρ. a time period of one year. no longer be in a position fully to meet its commitments to a bank. either to meet its commitments to the bank in full. Unless otherwise speciﬁed. Any borrower may be in a position.3) (2.6) (2. we will always be considering.4) B = breakdown distribution probability value b = breakdown distribution rate probability The breakdown distribution rate is in turn a probability dimension. in which i represents the interest that has accumulated: b= B L · (1 + i) (2.5) The expectation value of loss in cases of bankruptcy amounts therefore to (1 − b) L. For any borrower and period of time the following deﬁnition therefore holds good: ρ+χ =1 (2. χ is thus deﬁned as the probability of any borrower being in a position. the credit shortfall risk ρ ∗ is as follows: ρ ∗ = (1 − b) · ρ 0 ≤ ρ∗ ≤ 1 ρ ∗ = Credit shortfall risk (2. The correlation between these three values is derived later in this chapter. ρ is therefore deﬁned as the probability that a borrower will.2) (2. It is therefore worth: 0≤b≤1 (2. we will always be considering a time period of one year. χ is therefore assumed to have values between: 0≤χ ≤1 χ = Survival chance It is worth noting that ρ and χ are not identical with Ψ j on the strength of these deﬁnitions (see Section 1. the chance χ may be deﬁned as the chance of any borrower being able to meet its commitments at any time within any future period of time. It may rather be that an expectation of a percentage recovery B may enter into the equation concerned.7) . Unless otherwise speciﬁed. ρ may therefore be assumed to have values within: 0≤ρ≤1 ρ = Shortfall risk On the other side of the coin. within some future period of time.

and so on: a factor of 2 is used at this point. for example. As it is normally unwieldy to make a presentation of orders of magnitude using decimal fractions. B. BBB. purely arithmetically. S. where the value of ρ ∗ lies between zero and one. 12 levels (AAA.8) (2.4 EXAMPLE OF A RATING SYSTEM IN TERMS OF FIGURES The recognised rating agencies subdivide the issuers of loans in such a way that a loan issuer is in principle no different from a borrower. The rating categories in this example are therefore deﬁned numerically. For the purposes of illustration.9) 2. DDD. level AAA must be selected more narrowly than level AA.e.10) The following range applies for κ on the basis of the range for ρ ∗ : 1≤κ≤∞ (2. DD. we are concerned here to capture the shortfall risk of a borrower in terms of a numerical probability. CCC. and so on. 2. and with it the ﬁnancial standing of a borrower. in line with the following principles (all sorts of other principles could of course be imagined!): 1. AA. The rating categories are in this sense deﬁned qualitatively as. 14/15].12) .Rating System 21 Following on from equation (2. As good borrowers in terms of credit-worthiness are substantially more price sensitive than bad ones. can be expressed in ﬁgures by ρ ∗ . In contrast to this.3 DEFINING THE CREDIT-WORTHINESS KEY FIGURE Under this deﬁnition the credit shortfall risk. we make the additional deﬁnition here of the credit-worthiness key ﬁgure κ as the inverse of ρ ∗ : κ= 1 ρ∗ (2. CC. in the case of Moody’s [MOOD90. the following relative widths ensue for the individual levels: AAA = 1 AA = 2 A=4 BBB = 8 BB = 16 B = 32 CCC = 64 CC = 128 C = 256 11 DDD = 512 DD = 1024 D = 2048 The total of these relative widths comes to: 2 j = 4095 j=0 (2. D). level AA is twice as wide as level AAA. BB. The possible values of 0 ≤ ρ ∗ ≤ 1 are thus assigned to rating categories.3) the value of χ ∗ is deﬁned. i. such a rating system is developed at this point. C. Supported by the above principles. A. as follows: χ ∗ = 1 − ρ∗ 0≤χ ≤1 χ ∗ = survival chance regarding the breakdown distribution rate ∗ (2.11) 2.

1 Rating system used in the context of this book Value of ρ ∗ (%) Rating AAA AA A BBB BB B CCC CC C DDD DD D from 0.9817 49.0244 0.4786 24.0733 0.7570 1.1.3663 0.1709 1.1013 6.13) This now results in the numerical rating system shown in Table 2.1709 0.1709 1.9878 100 ρ ∗ according to rating level (%) 0.3663 0.2.3663 0.5385 3.2271 to 0.2271 12.5031% DD = 25.5385 3.1954% CCC = 1.7814% C = 6.0000 0. A simpliﬁed rating system for less competitive markets may be drawn up.0244 0.2271 12.2271 100 κ according to rating level (rounded) 585 65 8 1 .9878 to 0.5385 3.0061% D = 50.4786 24.1258% B = 0.1013 6.9817 49.2271 100 ρ ∗ according to rating level (%) 0.0733 0.2 Simpliﬁed rating system Value of ρ ∗ (%) Rating A B C D from 0.7570 1.0977% BBB = 0.0000 0.0244 0.9817 49. with only four levels as shown in Table 2.2515% and the following applies: 11 j=0 DDD = 12.1709 0. derived from Table 2.7570 1.22 Risk-adjusted Lending Conditions The following effective level widths ensue: AAA = 1/4095 = 0.9878 100 κ according to rating level (rounded) 4095 1365 585 273 132 65 32 16 8 4 2 1 Table 2.1709 0.3907% CC = 3.0244% AA = 2/4095 = 0. Table 2.1013 6.5585 6.5585 6.1709 1.2271 12.1.0733 0.4786 24.0122% 2j =1 4095 (2.5629% BB = 0.0488% A = 4/4095 = 0.5585 6.

C = 64 and D = 512. however.1709 0.Rating System 23 The relative widths of the individual rating levels here come to A = 1.0244 0. above all.1709 0. By analogy with the method of calculation in the preceding section. and for the AAA-. It should therefore be pointed out that the system presented in Table 2. AA. Table 2. be superﬂuous.3 could be developed over all levels to D. A and BBB ratings in Table 2.0063 0.1 may be further reﬁned.1 might therefore be insufﬁciently precise for such markets.0986 0.1709 0.0371 0. There are three times fewer levels in the simpliﬁed system in Table 2. The factor from level to level thus amounts to 23 = 8. being the third root of 2.2857 0.0986 0. It must be noted that ρ ∗ and κ are in each case identical for the AAA.3.2 than in the system in Table 2. B = 8.1306 0.0143 0.5 AMPLIFICATION OF THE RATING SYSTEM FOR VERY COMPETITIVE MARKETS The markets for loans to debtors with very high ﬁnancial standing.and BBB.0143 0.3663 ρ ∗ according to rating level (%) 0.0063 0.2217 0.0733 0. Analogously.0986 0.0733 0.0244 0.1306 0.0531 0. as the markets for loans to borrowers of lower ﬁnancial standing become progressively less competitive.2. are often extremely competitive. A.1.2217 0.3663 κ according to rating level (rounded) 15755 6971 4095 2694 1883 1365 1014 765 585 451 350 273 . as three times fewer levels occur than in the preceding example.1306 0. the exponent 3 being attributable to the number of levels being three times smaller.0371 0. This would.0531 0. 2.3 when the number of levels is tripled.0000 0. this factor comes to 21/3 .0244 0. AA-.0733 0. Table 2.0531 0.0063 0. one thus obtains the reﬁned rating system shown in Table 2.2217 0. in the case of a system with three times as many levels.ratings in Table 2.2857 to 0.3 Reﬁned rating system Value of ρ ∗ (%) Rating AAA+ AAA∗ AAA− AA+ AA∗ AA− A+ A∗ A− BBB+ BBB∗ BBB− from 0.0143 0. The relative width of the individual levels therefore grows by a factor of 8 = 23 . The rating system presented in Table 2.2857 0.0371 0.

.

Part II Mathematical Foundations of the Model Probability model: Development of ψ j Calculation of the shortfall risk hedging rate in the special case of shortfall risks being constant Calculation of the shortfall risk hedging rate in the general case of variable shortfall risk Shortfall risk on uncovered loans on the basis of statistics .

.

S.3. r ϕ(n) is the probability of the borrower defaulting at some point during the term of the loan r ε(n) is the probability of the borrower not defaulting during the whole of the term of the loan of n periods.2 we will show how the shortfall risk and survival chance might be converted over various terms.3 Probability Model: Development of ψ j As may be inferred from the basic equation (1.1) (3. 312]): ρj + χj = 1 ϕ(n) + ε(n) = 1 The results are derived in Appendix 4. For the sake of clarity the results of Chapter 3 will be presented again in Section 3. r χ j is the probability of the borrower not defaulting within period j. r j represents the period concerned: 1 ≤ j ≤ n. of n periods.2) ψj = εj = k=1 χk = k=1 (1 − ρk ) (3. determining the probability ψ j of cash ﬂow C j being fulﬁlled is of decisive importance for the model we are describing. r ϕ j is the probability of the borrower defaulting between the ﬁrst period and period j. The components needed for this model are deﬁned as follows (with the verb ‘to default’ being used as a synonym for the sentence ‘to no longer be able to meet commitments to the bank in full’): r n is the number of periods in the term of the loan. The correlation between the shortfall risk ρ and the survival chance χ and of the probability of fulﬁlment ψ j will be derived in Section 3.2).1 for loans that are unlimited in time and for ‘reasonable’ terms in relation to the shortfall risk ρ.1 DETERMINING THE PROBABILITY OF CASH FLOWS BEING FULFILLED What we are concerned to do below is develop a model for determining probabilities using equation (1. r ρ j is the probability of the borrower defaulting within period j.18) . They are: j j (3.3)) (cf.1. Conclusions may be drawn from the results of Section 3. 3. [BOHL92. This will be presented in Section 3. with the aid of probability calculus.4. The following correlations apply by deﬁnition (see also equation (2. In Section 3. r ε j is the probability of the borrower not defaulting between the ﬁrst period and period j.2).

the following applies: ψ j = χ j = (1 − ρ) j if χ1 = · · · = χn = χ (3. one quarter (ρq . a period of six months (ρs . . We can thus now detail the following maturity transformations by using the analogy for the special case ρm1 = · · · = ρm12 = ρm and χm1 = · · · = χm12 = χm .17) (see Appendix 4): 12 12 ϕ(12M) = 1 − j=1 χm j = 1 − (1 − ρm j ) j=1 (3. according to equation (3.25) (3. χm ).19) 3.30) (3. χq ).14) (see Appendix 4): 12 ε(12M) = j=1 χm j (3. however.31) 1 − ρ y = 1 − (1 − ρq ) = 1 − (1 − ρm ) 2 6 1 − ρy = 1 − 1 − ρy = 1 − χs2 1 − ρs = 1 − (1 − ρm )3 6 12 1 − ρs = 1 − 12 χm 3 1 − ρq = = χy = √ 2 6 χs = χ y = χq = χm √ √ 3 χq = 4 χ y = χs = χm √ √ √ χm = 12 χ y = 6 χs = 3 χq 4 χq .2 MATURITY TRANSFORMATION What we are concerned to prove now is that the values of the probabilities ρ and χ under the above deﬁnitions. ρ y = 1 − (1 − ρs )2 = 1 − (1 − ρq )4 = 1 − (1 − ρm )12 ρs = 1 − ρq = 1 − ρm = 1 − 4 (3. Let us assume a loan with a term of 12 months. χ y ).29) (3.28) (3. are dependent on the length of the period.22) and according to equation (3.21) By analogy. Then the following applies in general. χs ). the following applies.28 Risk-adjusted Lending Conditions In the special case of shortfall risks being constant. in relation to the same borrower. .26) (3.15) (see Appendix 4) in special cases: 12 ε(12M) = χm if χm1 = · · · = χm12 = χm (3. according to equation (3.24) (3. and to one year (ρ y . We will therefore in this section distinguish between the values for ρ and χ in relation to one month (ρm . . ϕ(12M) is the same as ϕ y .13) (see Appendix 4): 12 ϕ(12M) = 1 − χm = 1 − (1 − ρm )12 if ρm1 = · · · = ρm12 = ρm (3. .23) On the basis of the deﬁnition.20) and in the special case using equation (3. and ε(12M) is the same as ε y .27) (3.

only apply for ρ > 0. otherwise its shortfall risk ρ would be precisely zero! From this it follows that every loan must be limited in time. The following applies under equation (3. some correlations may be shown that ensue as a direct consequence of the properties of geometric series.13) (see Appendix 4) for the case where ρ = 0: ϕ(∞) = 1 − (1 − 0)∞ = 1 − 1∞ = 0 if ρ = 0 (3.1 ε(n) is given for various values of ρ.34) .3.Probability Model 29 Further maturity transformations may be worked out analogously.15) as an example (see Appendix 4). It follows from the illustration that all ε(1/ρ) for ρ → 0 have about the same value. In the course of operational banking this leads to the need to review every loan. under this procedure. from time to time.32) does. however. because a zero/zero division otherwise arises.12) is reformulated as follows for an inﬁnite number of periods: ϕ(∞) = ∞ j=1 (1 − ρ)( j−1) · ρ = ρ · ∞ j=1 (1 − ρ)( j−1) if ρ > 0! (3.1 The Case of a Loan being Granted Indeﬁnitely Under Swiss banking practice. The maximum ε(1/ρ) for ρ → 0 may be calculated using equation (3. as borrowers with shortfall risks of zero do not exist! In the case of overdraft facilities. or possible notice. 3.2 Reﬂections on the Success Chance ε(n) In Figure 3.33) From this one may draw the conclusion that every borrower defaults at some point if the time span for which credit is granted has an inﬁnite number of periods. as under normal contract clauses notice to terminate the facility may be given at any time.3. however. change in the frequency of review. this requirement becomes relative. 3. Equation (3. n = 1/ρ is a characteristic number of periods in this for each curve. 3. If there are changes in the risk position. ρ→0 lim ε(1/ρ) = lim (1 − ρ)1/ρ ρ→0 (3. current account credit in particular often has no time limit set on it contractually. a logarithmic scale having been selected for the n axis. then appropriate action must be taken: adjustment of the rate of interest.32) 1 1 =ρ· =1 ϕ(∞) = ρ · 1 − (1 − ρ) ρ Equation (3.3 CONCLUSIONS Using the special case ρ1 = · · · = ρn = ρ as an example. according to its shortfall risks.

Conversely. or conversely. it reduces to 45 millionths in the case of any term being 10 times as long.1 The substitution x = 1/ρ leads to [DMK/DPK92. for n = 1/(ν · ρ) lim 1 ν·ρ = lim (1 − ρ) ν·ρ = lim ρ→0 1 ρ→0 x→∞ 1− 1 ν·x x = 1 eν (3. By introducing the ratio ν.35) 1 ≈ 0.30 Risk-adjusted Lending Conditions (n ) (n = 1/nr ) 100% 100% n = 10 ρ = 10% ρ = 1% ρ = 0. It looks better if shorter periods of time related to n = 1/ρ are considered.8% 0% 1 10 100 1000 10000 0% n Figure 3.1.8%.5% n=2 50% 60.5% in the case of any term being 10 times as short. 33] lim ε(1/ρ) = lim ε(x) = lim x→∞ ρ→0 x→∞ 1− 1 x x = 1 e (3.7% 50% n=1 36. or to practically zero! So that the .1% 90. S.36) This leads for various values of ν to the following values for ε(1/(ν · ρ))max and ϕ(1/(ν · ρ))min : As may be inferred from Table 3.8% e which means that a loan with term of 1/ρ periods has a maximum probability of success of about 36. analogously to the above. the maximum probability of the success of the loan improves to 90. a minimum probability of loss of about 63.368 = 36.2%. the following may be set out.

That results in 1/(ν · ρ) = 0. all loans should be limited in time or provided with a contractual clause providing for unilateral termination.1 and in turn in subsection 3.24) (3. then the values for t periods may be calculated as follows. According to Table 3.5.2. This follows from the fact that any loan with an inﬁnite term will. this loan must be assessed every six months.0% 1.e. It is intended to clarify this by an example: let us assume that the risk of default on a loan should not be greater than 0.18) and.2% per annum.1 this corresponds to a value of ν = 1000.5% 9. 3. the shortfall risk ρ and the survival chance χ : j j ψj = εj = k=1 χk = k=1 (1 − ρk ) (3.2 form the necessary aids to decision making in this respect.3% 10 90. A bank’s lending policy will ﬁnd its expression in the value ascribed to this factor. in accordance with practical considerations.1% 31 shortfall risk of a loan may be kept ‘small’.2 results ν emax jmin 0. .2% 2 60.0% 1000 99. the number of periods in the term of the loan must be ‘small’ in relation to the quotient 1/ρ. The shortfall risk might be ρ = 0. respectively: ψ j = χ j = (1 − ρ) j if χ1 = · · · = χn = χ (3. default at some point in time.1% up to the next loan review. t may be assumed to have all values between zero and inﬁnity: ρ(t) = 1 − (1 − ρ)t χ (t) = χ t (3. this allows us to derive a policy on the intervals of time at which loans should be examined.8% 63.3.1 Subsection 3.28) Under the conclusions of Section 3.Probability Model Table 3. The methods of calculation outlined in Table 3. the following correlations exist between the probability ψ j . Seen from the operational banking point of view. i.7% 39.5% 100 99.1 45ppm ∼100% 1 36.3. The maximum term must be selected in proportion to the quotient 1/ρ.9% 0.4 RESULTS AND CONCLUSIONS As we have managed to show in Section 3.19) In the event that the shortfall risk ρ or the survival chance χ are known for a speciﬁed period of time.3.

.

4) and (3.1 FIXED ADVANCE WITHOUT REPAYMENTS Loans of this kind take the form.4) and (3.11). It is completely clear that the assumption of the shortfall risk ρ being constant over the entire term of any loan is unrealistic. and that assessment of loans follows the same rules. using equations (1. considering for the time being just the special case ρ1 = · · · = ρn = ρ over the entirety of n periods of the loan term. (2.4 that all known forms of clean credit may be described substantially with the same equations as in Section 4. That shortfall risks are not in fact constant over time will be shown in Section 7.4 Calculation of the Shortfall Risk Hedging Rate in the Special Case of Shortfall Risks being Constant This chapter is concerned with calculating the shortfall risk hedging rate ρ for different types of loan.7 summarises the most important results again. the breakdown distribution rate probability b. 4.6 are central to this chapter. The following thus applies.1. Based on Section 4. the total of the loan is paid out. the shortfall risk ρ and the risk exposure ρ ∗ will be derived in Section 4. covering the regular interest payments and the repayment of the loan at the end of the term.1 and 4.e. at ﬁrst. If the loan is covered.5 (cf. The correlation between the shortfall risk hedging rate r . Section 4. Figures 7.1. This assumption will be dropped in Chapter 5.19). and we will show how the general case of shortfall risks being non-constant may be combined with the results of Chapter 4.8 by means of an example. some operational conclusions are derived in Section 4.1. the equivalent values should count as being included in the collateral.2 to 4. This course of action was chosen in order to make the derivation of the results more open.2). We will show in Sections 4. There then follow several cash ﬂows from the borrower to the bank. according to equations (1. Sections 4. of a cash ﬂow from the bank to the borrower. (2. i. and these are illustrated in Section 4. the breakdown distributions concerned take the place of repayment.6.1) Λ : loan market value .10 and 7.19): Λ= χj ·i ·L j j=1 (1 + i s ) n + χn · L + (1 + i s )n n j=1 χ j−1 · ρ · b · L · (1 + i) (1 + i s ) j (4. In the event that the borrower defaults.2).

34 Risk-adjusted Lending Conditions L represents the amount of loan paid out. it is a consequence of the properties of geometrical series. The bank would otherwise be already accepting a loss at that point. The ﬁrst summand (in the ﬁrst set of brackets) represents the sum of the discounted expectation values of the interest payments. If a 100% breakdown distribution were assumed in an extreme case.1 and 4. The result reads as follows (see Appendix 5 for derivation): r= ρ∗ · (1 + i s ) 1 − ρ∗ (4. The higher the risk-free rate of interest i s . It is thus being assumed. Here it has been assumed that the breakdown distributions will be paid out at the end of the period in which the borrower defaulted and will always be of the same size. In what follows it is intended that r is calculated in such a way that the market value and nominal value of the loan are identical when it is paid out: Λ = L (cf. irrespective of the period concerned. risk-free interest and shortfall risk.14) In this way the ﬁnal result for the shortfall risk hedging rate r is. The second summand represents the expectation value of the discounted loan repayment. however. its market value must correspond at least to its nominal value: Λ ≥ L. for the sake of simplicity and without departing much from reallife circumstances. . L(1 + i) represents the lender’s demands. The factor ρ represents the probability that default on the loan does occur in period j. When a loan is paid out. Section 1. and with it the risk premium too. They do. The correlation between r . The third summand (in the second set of brackets) represents the sum of the expectation values of the breakdown distributions for each individual period. they did.6).2. following from the assumption that shortfall risks are constant. Their starting equation is therefore very similar. Seen mathematically. The factor χ j−1 represents the probability that default on the loan will not have occurred in the ﬁrst j − 1 periods. then the risk of default on the loan would be zero. i s . also have to establish a correlation between loan interest. as interest payments do indeed also have to be ‘insured’ from the point of view of risk. however. Under our assumptions interest was in practice paid during the ﬁrst j − 1 periods. deal with the general case right from the beginning (see later in Chapter 5). independent of the number of periods in the term of the loan! This comes about because the revenue arising from the shortfall risk hedging rate in the case of interest being paid in each period precisely covers the rise in the shortfall risk of the loan repayment. Roberts and Skimmer is interesting at this juncture. and ρ ∗ and b is portrayed graphically in Figures 4. where for the sake of simplicity it is assumed that the full amount of interest for the period j will become due in the case of bankruptcy. Comparison with the study of Fooladi. the higher will also be the risk premium r in the case of the same credit shortfall risk ρ ∗ . Their thesis is indeed the duration of obligations in cases where credit-worthiness is at risk [FRSK97]. that if bankruptcy occurs at all it occurs precisely at the end of the period concerned.

2 .1 r 1% 0.5% 0% 0% 50% r = 1%.Constant Shortfall Risk 35 Figure 4. is = 3% b 100% Figure 4.

The following problems for the bank in respect of the three elements that make up the loan interest rate (see equation (1. The same considerations apply analogously to this type of credit as to ﬁxed advance with regular repayments. when the credit is granted but not utilised. as the proportion of interest in the total annual payment becomes smaller and smaller owing to the repayments accumulating (cf. as in this case there are indeed costs. The worst case is.36 Risk-adjusted Lending Conditions 4. and thus the highest contributions to the covering of costs. The best case for the bank is the permanent and full utilisation of the credit. Equation (4. in that each individual repayment tranche is regarded as an independent loan in the form of a ﬁxed advance disbursement. As the amount is ﬁxed. quarterly or every six months) to the lender or lessor. Towards the end of the term it is exactly the other way round. as this allows the highest amounts of interest. Loan interest is only calculated on the amount of credit actually utilised each day. consumer credit and leasing credit is mainly granted in the form of loans on regular annual repayment. 39–41]. alter the validity of equation (4. Proﬁt Contribution Rate p Granting and supervising current account credit creates costs for the bank which are incurred anyway. it contains a high proportion of interest and a low proportion of repayment at the beginning of the term. This does not. on account of their independence from the number of periods in the term of the loan.4 CURRENT ACCOUNT CREDIT In the case of current account credit.2 FIXED ADVANCE WITH REGULAR REPAYMENTS This type of loan relies on ﬁxed advance disbursement without repayments. The only difference between these two types of credit lies in the fact that the sizes of the ‘imagined’ part loans vary considerably. The amount includes both interest and repayment. the bank agrees a maximum credit limit with the borrower. whether or not the credit is then taken up. 4. however. without repayments. 4. but no service is .14) applies again to these individual loans. [BRMY96. and therefore for the loan as a whole too.14). Financing Rate f No problems arise here if we make the assumption that only those costs which are incurred in proportion to the amount of credit taken up are built into the ﬁnancing rate f . vice versa. to be billed. S.3 LOANS ON REGULAR ANNUAL REPAYMENT In Swiss banking practice. The borrower or lessee thus makes regular ﬁxed payments (for example monthly.1)) arise from this. In contrast to ﬁxed advance credit this gives the borrower leeway to decide each day how much credit to take up. The same shortfall risk hedging rate results for all the part loans. as he draws money in or pays it out. as the case may be.

agreed with the customer. The overall necessary return from the shortfall risk hedging rate Rn across a bank’s entire current account credit portfolio of m individual credit arrangements is calculated by the summation of all m individual loans at the shortfall risk hedging rate ρ ∗ concerned. Traditionally the Swiss lending market has. Banking practice does.14). multiplied by the average amounts of credit taken up in each case L u . It is intended to assume in what follows. multiplied by the loan granted. it is not our intention to go into this problem more closely here. Experience shows that current account credit is fully utilised at the time of default. which is why the solution given below has been developed.20) Rb = billed return rcac = current account credit write-off risk hedging rate L u = used loan The simplest solution for bridging this difference consists in demanding an arrangement fee for each current account credit.14). however. utilisation is by no means always necessarily full prior to default. by virtue of the very nature of current account credit.14) for all borrowers by .19) Rn = necessary return L g = granted loan But only the current account credit write-off risk hedging rate rcacj . Revenue from the shortfall risk hedging rate only arises if the credit is taken up. The size of this arrangement fee corresponds here to the shortfall risk hedging rate under equation (4.14) may not be applied across the board in the case of current account credit. a full risk premium. however. Shortfall Risk Hedging Rate r The situation here is similar to the proﬁt contribution rate. however. So no problem arises as long as the average utilisation of the loan prior to default on it is the same as the extent to which the credit is taken up at the time of the default. paint a different picture. risk for the bank only arises for the bank if it is taken up and only to the extent that it is utilised. whether or not the credit is taken up. that in the case of current account borrowers the bank multiplies the risk rate under equation (4. not permitted this way of proceeding. and from time to time is even actually exceeded. however. On the other hand. may be billed: n Rb = j=1 rcacj · L u j (4. The circumstances outlined have the effect that equation (4. for the sake of simplicity. Even borrowers who do not default do not always use their credit limit absolutely fully and thus never pay. according to equation (4. In accordance with Section 1. Vice versa.2. in relation to the limit granted (which corresponds de facto to the bank’s effective lending risk). multiplied by the loan granted j in each case L g (see Appendix 5 for derivation): m Rn = j=1 ρ∗ j · (1 + i s j ) · L g j 1 − ρ∗ j (4.Constant Shortfall Risk 37 provided that can be billed.

allows us to come to grips with this ∗ difﬁculty more successfully.24) L uk j j=1 which is reduced to: mk L gk j rck = j=1 mk (4. In times of rapid economic change this might only be possible very imprecisely.20) and (4. however. (4. Introducing a rating system. as in this case the amounts of credit utilised L u j have to be estimated for the future. as explained in Chapter 2.22) As the correction factor is a constant. with the aid of equations (4.21): m j=1 ρ∗ j 1 − ρ∗ j · (1 + i s j ) (4. The ﬁgures may therefore be taken out of their brackets as follows: ρ∗ j 1−ρ ∗ j ρ∗ j 1−ρ ∗ j · (1 + i s ) · · (1 + i s ) · mk L gk j j=1 mk rck = (4. The imputed ρk for all loans within a deﬁned rating level k are identical and therefore constant.21) ρ∗ j 1 − ρ∗ j m · (1 + i s ) · L g j = j=1 rc · ρ∗ j 1 − ρ∗ j · (1 + i s ) · L u j (4. The following equation for determining rc may now be drawn up. Subsequent division gives the following result for rc : m ρ∗ j 1−ρ ∗ j ρ∗ j 1−ρ ∗ j · (1 + i s ) · L g j (4. it can be placed in front of the sum.19). The calculation of rc for future periods is. though only with considerable computation. the intention is to achieve a position in which the risk premium that is necessary and the risk premium that has been billed are identical. more problematic.38 Risk-adjusted Lending Conditions a correction factor rc : rcacj = rc · rc = correction factor Using this correction factor. the resulting calculation can be undertaken without further ado.25) L uk j j=1 The risk premium rate rcack for any current account credit at the rating level k with m k ∗ individual loans and the credit shortfall risks therefore associated with them of ρk is thus .23) · (1 + i s ) · L u j rc = j=1 m j=1 In relation to the past.

As invoices for works completed are in each case settled as late as possible. because of the quotient of the two totals in equation (4. however. The problem of ‘inventory limits’ is at its most conspicuous with construction loans.26) j=1 The values for the future periods to be calculated must be inserted in the case of both totals in equation (4. if possible. Furthermore the difﬁculty of ‘inventory limits’ could be effectively countered. If a construction loan deteriorates from the point of view of credit-worthiness during the construction period — whether it be owing to the client’s ﬁnancial standing or to the quality of the construction project in hand — it is usually pointless for the bank to call in the loan. be a means of canvassing new mortgage business.e. seen in isolation. This results in so-called ‘inventory limits’ (i. Attractively priced construction loans may. the less need it has to bring imponderables into its considerations. or whether a special procedure has to be developed. in the context of being competitive as far as interest rates are concerned. a competitor has a similar credit limit at its disposal. longer term unutilised current account credit limits) inevitably raising the level of current account interest rates.26) this leads to a shortfall risk charge three times higher than that for full utilisation. then on the other hand it is still only the price element f that . By their very nature these are only utilised fully at the end of the term. High unutilised current account credit limits lead to high rates of interest on current accounts. If the price elements p and r (see equation (1. may also be solved by arrangement fees on credit limits granted. For the borrower this means that it should be budgeting its future credit needs as precisely as possible and applying to its bank for an appropriate credit limit. the extent to which any construction loan is taken up is equivalent to an average utilisation of about one-third. i. have historically been under-priced. We permit ourselves at this point the conjecture that construction loans. as during the term of such a loan there is only an unﬁnished building to show for it. to pay the loan off. ‘Inventory limits’ are thus to be avoided.Constant Shortfall Risk 39 calculated in the following way: mk L gk j rcack = j=1 mk · L uk j ρ∗ j 1 − ρ∗ j · (1 + i s ) (4. In practice it remains to be investigated whether the future values may be approximated sufﬁciently precisely by reference to past values. and therefore be justiﬁed on marketing grounds. as presumably only a very few borrowers are prepared to pay an arrangement fee for a credit limit that may hardly ever be utilised. The more accurately it is in a position to budget. or because the borrower wishes to make capital out of the size of its credit limit.e.26). This would also be a solution in relation to the billing of proﬁt contributions that might not otherwise be obtained. but this presupposes a matching degree of customer loyalty. Under equation (4. This has the consequence that borrowers should not apply to banks for unnecessarily high credit limits just because. In this situation the bank has no alternative but to complete the building at its own expense and therefore in practice.1)) are billed in the form of an arrangement fee. de facto. particularly in the case of current account loans. As already mentioned the problem of only partial utilisation.26). for example. prior to consolidation.

usually also an indication of deteriorating creditworthiness on the part of the borrower. however. which would cover the additional risks even in the most unfavourable instance. one may always fall back on current account lending. 4. One further problem in current account lending is the tiresome subject of current account facilities being exceeded. Any borrower’s credit-worthiness may. and calls therefore for reassessment of the lending. lead to a current account credit priced today with interest at 5% per annum and with a fee of 0. of three or six months. as experience shows that current account credit facilities are fully utilised in such situations and therefore are tantamount to de facto ﬁxed advances. purely qualitatively. The exceeding of credit limits is. being charged in future at 4% per annum on the amount of loan actually taken up and at a fee of 0. As the allowing of credit to be exceeded means a great and in no way automatic concession by the bank. be forced upon the Swiss market for credit at the moment. a decision must be taken as to whether the normal credit limit could be raised. however. providing the risk-free rate of interest i s is identical in both cases. Moreover the qualifying time until maturity in the case of ﬁxed advances.25% per quarter being demanded. This is no arithmetical problem here. The most effective method of meeting this difﬁculty is the charging of fees or of a special rate of interest on the excess. for example. that any current account credit facility for the same borrower is always associated with a higher loan interest rate than ﬁxed advance credit. If winding up such advances does occur. tend to be more expensive than ﬁxed advance credit. Here it must also be established.28) (4. matching prices may also be applied here. If cases of exceeding limits are ignored. change during the term of . would not be excessively long. whether or not the high loan interest rates associated with the excess borrowing can be borne by the borrower: over-mechanical applications of higher rates of interest may cause additional credit-worthiness problems and thereby become self-fulﬁlling prophecies. or whether it is a case for withdrawal of credit.40 Risk-adjusted Lending Conditions rises in proportion to the loan being taken up. in particular.27) This allows us to note. The advantage of being able to give immediate notice on current account facilities in cases indicating that those facilities should be withdrawn is in many cases illusory anyway. Such an arrangement fee should not. So a lending policy could consist in only granting ﬁxed advances that are fully utilised at any one time and operating current accounts only on the basis of credit balances. But even if i s is not identical. In practice this may. it may be said that the following always applies: L uk < L gk From this it may be concluded that the following always applies: rck > 1 (4. then it may be emphasised that current account credit must.5% per quarter on the amount of credit granted. If credit limits are exceeded over a longer period of time.5 LOAN ASSESSMENT This chapter has so far shown how the shortfall risk hedging rate r should be calculated when paying out a loan. on the strength of the expositions in this section. however.

i.45) (4.6). An appreciation proﬁt thus exists if the right-hand bracket in the numerator of equation (4.38) The left-hand bracket in the numerator and the denominator in equation (4. (1 + i) · (ρl · b + χl ) = 1 + i sl (1 + i) = i= i= 1 + i sl ρl · b + 1 − ρl (4.Constant Shortfall Risk 41 a loan.38) is positive.41) (4.38) are always positive because of the deﬁnitions of the values occurring.45) is identical to equation (4.29) The value of i sl at the time of the assessment does not necessarily have to coincide with the value of i s at the time of paying out the loan. λ= (1 + i sl )l − χll · ((1 + i) · (ρl · b + χl ) − i sl − 1) (1 + i sl )l · (1 + i sl − χl ) (4. There is thus no need for assessment correction if the risk-adjusted rate of interest il l periods prior to the end of the term is identical to the risk-adjusted rate of interest when the loan is paid out! 4. By applying equation (4. which is expressed by the index I .39) 1 + i sl −1 1 − ρl · (1 − b) 1 + i sl − 1 + ρl · (1 − b) 1 − ρl · (1 − b) i= i sl + ρl · (1 − b) 1 − ρl · (1 − b) i sl + ρl∗ = il 1 − ρl∗ i= Equation (4.49) (see Section 4. If it is negative.14). . The equation factor λ for correcting the assessment of market value in relation to nominal value is derived in Appendix 5. circumstances have arisen requiring a provision to be made: (1 + i) · (ρl · b + χl ) > 1 + i sl ⇒ appreciation proﬁt (1 + i) · (ρl · b + χl ) < 1 + i sl ⇒ requiring a provision No assessment correction is necessary if this bracket is equivalent to zero.6 CONCLUSIONS The following conclusions can be drawn by derivation from equation (4.43) (4.44) (4.e.42) (4.40) (4.1) the market value of a loan L l for the last l periods of the term can be detailed as follows: l Λl = j=1 χl · i · L (1 + i sl ) j j + L + (1 + i sl )l l j=1 χl j−1 · ρl · b · L · (1 + i) (1 + i sl ) j (4.

48) (4.46) can be simpliﬁed as follows: i s · (1 − ρ ∗ ) + ρ ∗ · (1 + i s ) i= 1 − ρ∗ i= is − is · ρ ∗ + ρ ∗ + is · ρ ∗ 1 − ρ∗ i= is + ρ ∗ 1 − ρ∗ (4.14) inserted into equation (1.46) (4.6.49) The loan interest rate under equation (4.50) See Figure 4.3) and (4.1 Minimum Loan Interest Rate Equations (1.42 Risk-adjusted Lending Conditions 4.49) can now be interpreted in the case of the minimum rate of interest given i smin or risk-free loans and the credit shortfall risk given ρ ∗ as minimum rate of interest i min .3 for illustration of the correlations. i Parameter: is 5% 1% 3% 20% 10% 0% 0% Figure 4.1) will yield: ρ∗ i = is + · (1 + i s ) 1 − ρ∗ Equation (4.3 10% 20% .47) (4. and it is this loan interest rate that must be billed to the customer in order to achieve the necessary minimum proﬁt contribution: i min = i smin + ρ ∗ f + pmin + ρ ∗ = 1 − ρ∗ 1 − ρ∗ (4.

This will become clear in the following subsection. when the value of i eff changes the value of peff does not change to the full extent of the value of i eff .6.60) (4.6. when i eff is so small. and vice versa.53) into use results in: reff = ρ∗ · (1 + f + i eff · (1 + ρ ∗ ) − f − ρ ∗ ) 1 − ρ∗ ρ∗ · ((1 − ρ ∗ ) + i eff · (1 − ρ ∗ )) 1 − ρ∗ reff = ρ ∗ · (1 + i eff ) Because ρ ∗ and i eff are always positive ﬁgures.53).50).57) (4.14): reff = ρ∗ · (1 + f + peff ) 1 − ρ∗ (4.50): i eff = f + peff + ρ ∗ 1 − ρ∗ (4.3 Effective Shortfall Risk Hedging Rate By analogy with the effective proﬁt contribution rate. in the case of a rise in the value of i eff .54) shows that. and indeed if that happens.2 Effective Proﬁt Contribution Rate The bank will in practice agree with the customer an effective loan interest rate i eff .51) (4. which is not necessarily identical to the minimum loan interest rate i min according to equation (4.59) reff = . the effective shortfall risk hedging rate can be derived by using equation (4. it is always the case that: reff > 0 (4.55) f + ρ∗ 1 − ρ∗ (4. peff may also assume negative values. this rise must also be covered again via a rise in the shortfall risk hedging rate. 4. which is thus not identical to the minimum proﬁt contribution rate pmin and can be derived as follows from equation (4.52) (4. but only by the reduced factor of 1 − ρ ∗ .56) Bringing equation (4.58) (4.Constant Shortfall Risk 43 4. This leads to an effective proﬁt contribution rate peff .53) i eff · (1 − ρ ∗ ) = f + peff + ρ ∗ peff = i eff · (1 − ρ ∗ ) − f − ρ∗ As is discernible from equation (4. That is because. that means: peff < 0 if i eff < Furthermore sensitivity analysis of peff versus i eff ∂ peff = 1 − ρ∗ ∂i eff (4.

using equation (4. in the case of any change in the value of i eff likewise does not change to the same extent as i eff .61) shows that the value of reff . Moreover it is that case that: ρmax = 0 if i max = f + pmin For the legislature laying down the value of i max . 4.67) makes it clear that higher values for i max permit higher values for ρmax too.49): i max = ∗ Solving.64) ∗ ρmax = i max − i smin i max − f − pmin = 1 + i max 1 + i max (4.62) shows. using ρmax .44 Risk-adjusted Lending Conditions The sensitivity analysis of reff in relation to i eff .65) ∗ Equation (4. but this time changes by the reduced factor ρ ∗ . . the essentially larger part of any change in the effective rate of interest on the effective proﬁt contribution ratio does not apply. The sum of both partial differentials equations (4.61) ∂ peff ∂reff + = 1 − ρ∗ + ρ∗ = 1 ∂i eff ∂i eff (4. The question thus arises — up to what maximum credit shortfall risk ρ ∗ may consumer credit be granted in the case of a given minimum risk-free standard rate of interest Ismin ? The following applies.63) As ρ ∗ is substantially smaller than (1 − ρ ∗ ) in the case of borrowers of good ﬁnancial standing. the necessity arises from this that: i max f + pmin (4. results in: ∗ i smin + ρmax ∗ 1 − ρmax (4. and indeed in the ratio peff 1 − ρ∗ = reff ρ∗ (4.4 Maximum Shortfall Risk Covered In some Swiss cantons the maximum loan interest rate i max for consumer credit is laid down by law. that any change in the value of i eff is distributed completely over the values of peff and reff . ∂reff = ρ∗ ∂i eff (4.55) and (4. as was to be expected.6.67) The market for consumer credit would otherwise be reduced to vanishing point owing to high credit-worthiness requirements.66) (4.

14) In the case of current account lending a correction factor rc must be introduced. Thus the effective proﬁt contribution rate and the effective shortfall risk hedging rate are calculated as follows: peff = i eff · (1 − ρ ∗ ) − f − ρ ∗ (4. The starting position looks like this: .59) There is an important observation here.8 EXAMPLE It is intended to grant a borrower a ﬁxed advance for one year. ever be negative! 4.Constant Shortfall Risk 45 4. however.21) We refer to the expositions in Section 4. This results in the effective loan interest rate i eff . it will in practice be rounded up to the next normal round ﬁgure (usually in the form of one-quarter.5 for calculation of the correction factor. It is particularly important here to note the numerator. The greater the shortfall risk hedging rate.2: r= ρ∗ · (1 + ir f ) 1 − ρ∗ (4. that the effective proﬁt contribution rate becomes negative under the condition: i eff < f + ρ∗ 1 − ρ∗ (4.54) The effective shortfall risk hedging rate may not. the more the numerator in equation (4. The minimum loan interest rate is calculated from the equation: i min = i smin + ρ ∗ 1 − ρ∗ (4.50) This is one of the most signiﬁcant equations.53) reff = ρ ∗ · (1 + i eff ) (4. one-eighth or one-sixteenth of 1%).49) makes itself noticeable: the minimum loan interest rate is not simply the sum of the minimum risk-free standard rate of interest and of risk! After the minimum rate of loan interest i min has been laid down for a loan transaction. depending on the average utilisation of the loan: rcacj = rc · ρ∗ j 1 − ρ∗ j · (1 + rscac ) (4.7 RESULTS AND CONCLUSIONS The shortfall risk hedging rate is normally calculated from the following equation according to Section 4.

7933% 1 − 0. according to equation (4.7939% = 4.04875 · (1 − 0.59).04) = 0.00757 = 1.49): 0.53) and (4.875% = 4 % 8 .00757 = 4.7933% 1 − 0.757% The minimum risk-free standard rate of interest is: i smin = 3% + 1% = 4% The shortfall risk hedging rate. as follows: peff = 0.00757 The ﬁgure of 4.7933% One obtains the same result.04 + 0.000% ρ ∗ = 0.875% 8 The effective proﬁt contribution rate and the effective shortfall risk hedging rate are calculated.7933% = 4.7939% The proof is: 7 i eff = 3% + 1.03 − 0.46 Risk-adjusted Lending Conditions Financing cost rate: Minimum proﬁt contribution rate: Credit shortfall risk: (Rating BB according to Table 2. moving directly.0811% + 0.1) f = 3.7933% is not customary in Swiss banking practice.04875) = 0. via the shortfall risk hedging rate r under equation (4.0811% reff = 0.1) by r= i min = 4% + 0. It is therefore rounded up to the nearest eighth: i min = 7 i eff = 4 % = 4.00757 From this is calculated the minimum loan interest rate according to equation (1. with the aid of equations (4.14) is: 0.00757 · (1 + 0.00757) − 0.000% pmin = 1.00757 · (1 + 0.

**5 Calculation of the Shortfall Risk Hedging Rate in the General Case of Variable Shortfall Risk
**

We now give up, in this chapter, the condition ρ1 = · · · = ρn = ρ. First the exact solution for the case of a ﬁxed interest loan without repayments will be derived in Section 5.1. In Section 5.2 we will draw up, using an approximate solution as an example, a comparison with the special case in Chapter 4, and we will examine the accuracy of the approximate solution in Section 5.3. Other important clean credits will be dealt with in Sections 5.4, 5.5 and 5.6. The important results and conclusions from Chapter 5 will be summarised in Section 5.7. The aim of this chapter is to elaborate general principles, which will then be developed further in Chapter 7.

**5.1 FIXED INTEREST LOAN WITHOUT REPAYMENTS
**

First it is intended to examine again, as in Chapter 4, the ﬁxed advance without repayments. Equation (3.18) must, however, now be brought into play for the value ψ j in equation (1.2). By analogy with equation (4.1) this results in (for cases where the loan is covered, the values are on the other hand valid when the collateral is included): j n i · L · (1 − ρk ) L · (1 − ρk ) n k=1 k=1 Λ= + (1 + i s )n j j=1 (1 + i s ) n b · L · (1 + i) · ρ j · (1 − ρk ) k=1 + j=1 (1 + i s ) j

j−1

(5.1)

The left-hand summand again corresponds to the discounted expectation values of the interest payments, the centre summand to the expectation value of the loan repayment and the righthand summand to the sum of the discounted expectation values of the breakdown distributions. It will again be assumed, as in Chapter 4, that there is one uniform risk-free rate of interest i s for all terms. Λ = L will again be set (see Section 4.1) for the period in which the loan is paid out.

48

Risk-adjusted Lending Conditions

**The following substitutions will be made in order to be make handling equation (5.1) more manageable:
**

j n

(1 − ρk ) (5.2)

x=

i=1 n

k=1

(1 + i s ) j (1 − ρk )

y=

n

k=1

(1 + i s )n ρj ·

j−1 k=1

(5.3)

(1 − ρu ) (5.4)

z=

j=1

(1 + i s ) j

With x, y and z brought into play, and abbreviated with L, there results: 1 = i · x + y + b · (1 + i) · z and solved using i: i= 1 − (y + b · z) x +b·z (5.6) (5.5)

Logically, no further simpliﬁcation of the representation can be achieved by any reverse substitution. Equations (5.2), (5.3), and (5.4) should therefore be seen as computational indications, the results of which may be brought into play in equation (5.6). These equations can be used with appropriate PC worksheets without any difﬁculty. The mean shortfall risk ρm , which implicitly underlies the interest rate calculated in this way, is calculated as follows using equation (4.67) as an example: ρm = i − is 1+i (5.7)

**5.2 APPROXIMATE SOLUTION FOR FIXED INTEREST LOAN WITHOUT REPAYMENTS
**

The computing needs for Section 5.1 are in practice indeed applicable, but very unwieldy. An approximate solution is therefore derived in this section. The shortfall risk ρk of the k th period can be replaced by the sum of the average shortfall risk ρa of all n periods and the k th deviation ρk by this means: ρk = ρa + ρk

n

ρk n

(5.8)

with ρa =

k=1

Variable Shortfall Risk

49

**Equations (5.2) for x, (5.3) for y and (5.4) for z now read:
**

j n

(1 − [ρa + ρk ]) (1 + i s ) j (5.9)

x=

j=1 n

k=1

(1 − [ρa + ρk ]) (1 + i s )n

j−1 k=1

y=

n

k=1

(5.10)

(ρa + ρ j ) ·

(1 − [ρa + ρk ]) (5.11)

z=

j=1

(1 + i s ) j

The approximate solution is derived in Appendix 6 using conventional approximation equation. The result is: ∗ i s + ρa (5.29) i= ∗ 1 − ρa Comparison with equation (4.49) shows that it is identical to equation (5.29). In the case of the variable ρ j equation (4.49) may thus be used, likewise approximately, in that the constant ∗ ρ ∗ is replaced by the average ρa . On the basis of the type and method of derivation used, this result could indeed have been expected.

**5.3 RELIABILITY OF THE APPROXIMATE SOLUTION
**

At this point the reliability of the approximate solution will be illustrated by means of six ∗ numerical examples. Let’s put i s = 5%, ρa = 1% and b = 0 in all the examples. The approximate solution is thus always i = 6.0606%. The values for the shortfall risk ρ concerned in the individual years may be taken from Table 5.1. The value of i, calculated precisely, is detailed in the third column from the right-hand side and may be compared with the approximate solution i = 6.0606%. The examples were selected as follows:

**r a and d represent deteriorating credit-worthiness r b and e represent improving credit-worthiness
**

Table 5.1 Reliability of the approximate solution Values for ρk year 1 a b c d e f 0.8% 1.2% 1.0% 0.6% 1.4% 1.0% 2 0.9% 1.1% 1.1% 0.8% 1.2% 0.8% 3 1.0% 1.0% 1.0% 1.0% 1.0% 1.0% 4 1.1% 0.9% 0.9% 1.2% 0.8% 1.2% 5 1.2% 0.8% 1.0% 1.4% 0.6% 1.0% i exact (%) 6.0481% 6.0733% 6.0631% 6.0359% 6.0863% 6.0557% Results abs. error 0.0125% −0.0127% −0.0025% 0.0247% −0.0257% 0.0049% rel. error 0.2062% −0.2092% −0.0418% 0.4093% −0.4215% 0.0816%

50

Risk-adjusted Lending Conditions

r c and f represent oscillating credit-worthiness r in c there is deterioration at ﬁrst and in f there is improvement at ﬁrst r in a, b and c the changes in each case are half the size of those in d, e and f.

The following can be seen from the results:

**r The approximate value for i is too high in the case of deteriorating credit-worthiness, and r In r r
**

too low in the case of improving credit-worthiness. the case of oscillating credit-worthiness it depends whether it has improved at ﬁrst (approximate solution too high) or deteriorated ﬁrst (approximate solution too low). The deviation of the approximate solution is about double in the case of changes that are double the size. Although great ﬂuctuations in credit-worthiness were inserted into the examples, the absolute error amounts to only a few basis points or fractions of a basis point.

As has already been emphasised in Section 5.1, it is possible to use the exact equations with appropriate PC worksheets. In most cases, however, usable results can also be delivered by approximate solution.

**5.4 FIXED ADVANCE WITH COMPLETE REPAYMENT
**

Equation (1.2) appears in this case as follows:

n

Λ=

j=1

i ·L · 1− ×

( j − 1) n

+

L n

·

j

(1 − ρk ) + b · ρ j · (1 + i) · L · 1 − (1 + i s ) j

k=1

( j − 1) n

·

j−1 k=1

(1 − ρk )

(5.30)

The ﬁrst summand in the numerator corresponds to the expectation values of the interest payments and repayments. The ﬁrst summand in the left-hand curved brackets corresponds to the interest payments, with the square brackets detailing the rest of the loan subject to interest in each case. The second summand in the left-hand curved brackets corresponds to the tranche of repayment. The second summand in the numerator corresponds to the expectation values of the breakdown distributions. The square brackets in the right-hand curved brackets represents the remaining debt. The other terms are analogous to those in equation (5.1). The result for i is derived in Appendix 6 on the assumption that Λ = L at the time of paying out. The result is: i= 1+

n j=1

n b · ρj 1−ρ j

· 1+

1 i

· (n − j + 1) + (1 + i s ) j

1 i

·

j

(5.38) (1 − ρk )

k=1

8994% 6.0137% −0.39) k=1 i s )n Here α represents the degree of repayment with 0 < α ≤ 1.0863% 6.0768% 5.0683% 0. d and f in the case of repayments.23) has to be supplemented as follows for this case: n L = j=1 i ·L · 1− × (1 − α) · L · + (1 + α · ( j − 1) n + α·L n · j (1 − ρk ) + b · ρ j · (1 + i) · L · 1 − (1 + i s ) j k=1 α · ( j − 1) n · j−1 k=1 (1 − ρk ) n (1 − ρk ) (5.0557% i with amo 5. The solution lies in each case between that for loans without repayments and that for loans with complete repayments over the term. It is now interesting to realise that lower interest rates result only in examples a.2234% 6.2 are calculated with the risks over term examples of Table 5. . the interest rates for ﬁxed advances with repayments in Table 5. Here repayments clearly have the effect of reducing risk.0284% i with less i without −0.0685% 0. because lower amounts of loan are bearing interest on account of repayments in periods of good credit-worthiness.1418% 6.0273% 51 The rate of interest i in equation (5.0359% 6. We will not go into the derivation at this point.Variable Shortfall Risk Table 5.1371% −0. In the cases of improving credit-worthiness it is the other way round.38) occurs both to the left and to the right of the equals sign.1 and compared with the rate of interest of ﬁxed advances without repayments.0733% 6. In these cases.1365% 0.9798% 6. 5. The standard rate of interest i s is again 5%. In this format the equation is therefore soluble iteratively with the aid of an appropriate PC worksheet. In order to illustrate equation (5.0481% 6. The solution path matches that in the preceding section and leads once more to an iterative solution. only correspondingly lower revenues may be billed. which leads to higher rates of interest in relation to loans without repayments. The result of this is that a higher margin is needed at the beginning of the term of the loan.5 FIXED ADVANCE WITH PARTIAL REPAYMENTS Equation (5.0631% 6.38).2 Interest rate comparison for loans with and without repayments Example a b c d e f i without amo 6. These are the cases of deteriorating credit-worthiness.

S. Chapter 6 outlines how to go about this. It also needs an operational accounting system. 5. Equations (4.1 and 5.29) form the necessary prerequisites for this. this chapter supplies the details needed to check the approximation or to calculate interest rates more precisely. As soon as the bank establishes such situations.8). it simply recalculates and adjusts the loan interest rate. Forecasts for the future can be drawn up supported by these and the corresponding values inserted into the model. in order to be able to calculate the parameters used in the model in relation to the past.2 show. The results to date are only applicable as an autonomous model if the bank has the necessary statistical data available. such as is normally found in insurance companies.49) and (5.52 Risk-adjusted Lending Conditions 5. real-life loans may normally be handled by using the average shortfall risk ρa according to equation (5. however. 673 and thereafter] in Chapter 7. in conjunction with the equations given in Chapter 4. Above all. .7 RESULTS AND CONCLUSIONS As Tables 5. there is no problem regarding changes in the borrower’s credit-worthiness. the calculations we have made so far serve as a basis on which to develop further the approach of Black and Scholes [BLSC73. If it is suspected that this approximation is insufﬁciently accurate.6 CURRENT ACCOUNT LOANS As interest rates on current account loans may be adjusted in the short term to new circumstances.

that invoices for medical services and credit card balances represent the most frequent cause of personal bankruptcy. both in terms of numbers and in terms of amounts.1 Unearned Income and Income from Self-employment The shortfall risk of the private individual in these cases corresponds to the shortfall risk of the pension fund or of his/her own company. Experience in Swiss bank lending shows that loans to businesses and to private individuals more or less balance out. etc. A development that may possibly be in store for Switzerland too?) Private individuals normally have four possible sources of income with which to ﬁnance their debt servicing: r income from salaried employment r income from self employment r pensions (old age and dependant’s insurance.) r investment income and assets Earned income and pensions are normally mutually exclusive and may thus be considered separately.6 Shortfall Risk on Uncovered Loans on the Basis of Statistics This chapter will open up opportunities for determining the shortfall risks of borrowers with the aid of statistics. Separate and detailed consideration of both these groups of customers is therefore justiﬁed. pension funds. The shortfall risk for loans to private clients on an uncovered basis will therefore be examined below.1 PRIVATE CLIENTS As will be shown in Chapter 8. 6. . The various sources of income and the effects they have on the shortfall risk will be examined separately below. It will therefore be referred to in Section 6. life insurance. (Domowitz and Sartain have as it happens established in a recent study for the USA [DOSA97]. when there is any. the shortfall risk of covered loans — as these predominantly are in the case of private client business (mortgages. Investment income comes. Any loan to a private individual is subject to default if the person concerned no longer has sufﬁcient income available to service the debt. on top of them. 6.1.2 instead. loans against other collateral) — is made up of the shortfall risk for uncovered loans to the same borrowers and of the shortfall risk of the cover.

This is obviously not easy. That would. as well as the personal qualiﬁcations of the borrower. and can be very precisely estimated on the basis of experience in the property sector. age and location. as indeed has already been done on occasion. however. The size of L s is thus aligned with the acceptability of the debt servicing potential. In the case of returns from property in private ownership.1. play a decisive role in this. This then becomes a situation for review. take the form of uncovered loans not being granted to private individuals in principle. and these may be assigned in favour of the loan. The risk for the bank consists in the possibility of the management of the property in question being improperly conducted. a ‘management agreement’ can be drawn up. We will not therefore go further into these risks at this point. but the bank nonetheless has to form a view for itself.54 Risk-adjusted Lending Conditions 6. From this the bank must then determine for itself whether this probability is more or less likely to befall the borrower concerned over the next year. be taken out with the insurance industry to cover these risks. Long-term and sustained revenue can be perfectly well achieved from real property. In addition to the risk of unemployment.1. The bank’s risk here is that the borrower becomes unemployed and remains so in the long term.3 Investment Income and Assets This subsection will distinguish on the one hand between securities and credit balance assets. Revenues from securities and credit balance assets are often very volatile.2. Loans that a bank may make against other collateral should be evaluated according to appropriate statistics. The case of income arising from returns on real property is quite different. and property assets on the other hand.2 Income from Salaried Employment It is assumed in this subsection that the bank will grant a private individual an uncovered loan amounting to L s on the strength of his/her salary. If a private borrower only has income from securities and credit balances available. then the shortfall risk should be calculated only on the basis of covering the loan. as will be described in Section 6. 6. Short-term unemployment can usually be bridged with the aid of unemployment insurance. . however. and of its long-term revenue thus being jeopardised. the risks of invalidity and death have to be taken into account. Such cases have a feature in common and comparable with that of companies. The probability of the borrower becoming unemployed in the long term within one year can be investigated on the basis of unemployment ﬁgures relating to his/her occupation. The probability of a person defaulting on an uncovered loan that is backed up solely by income from assets must therefore be considered very high. taking account of salary in accordance with the normal rules of consumer credit business. The size of the loan L s in relation to salary is determined on the basis of the bank’s credit policy. Appropriate policies may. also mean not granting any normal consumer credit either! Establishment of loan amount L s therefore takes place in line with appropriate practice. A defensive attitude by the bank may. Economic forecasts for the occupational or professional group concerned and for the region. in an extreme case.

in particular. have sufﬁciently accurate facts at their disposal.2 COMPANIES Historically orientated deﬁnitions of the probability of a company defaulting on a loan can be made on the basis of bankruptcy statistics. and which provide clues that will be applicable for the following period. based on the one hand on historical (but reappraised) facts and on the other hand on economic forecasts. Here it makes sense to weigh up against each other the need for the neatest possible classiﬁcation and the need for results that are statistically meaningful. on the other hand.Shortfall Risk on Uncovered Loans Using Statistics 55 6. . and on the need to obtain statistically meaningful results. How far back the necessary period for consideration should go depends on the size of the individual segments. The most intractable problem is certainly classiﬁcation by sectors. and the cantonal banks within their cantonal areas. The facts behind the bankruptcy statistics should be as up to date as possible for the purposes described above. We will not go any further into such statistical methods here. should. as many companies clearly cannot simply be ﬁtted into just one particular sector. In order to arrive at the most precise ﬁgures possible for individual companies it is necessary to classify the entirety of all companies as neatly as possible. On the other hand the observation period should not be too long. the future is far from simply being an extrapolation of it. even if they only use them for making evaluations within their own established clientele. Analysing bankruptcy statistics always involves reappraisal of the past but. On the strength of their high market shares the major banks. The degree of relevance required is not necessarily provided by facts that are made available from ofﬁcial sources. however. Banks cannot therefore avoid forming an opinion on the risks of defaults arising in individual segments over the coming periods concerned. Particular criteria for classiﬁcation are: r business sectors r geographical/economic areas r sizes of enterprises by number(s) of employees r how long companies have been established Such details may be obtained — for foreign companies too — and are included in every census of business operations in Switzerland. in order to have facts that are as up to date as possible. We refer you at this point to appropriate specialist reading (such as [BOHL92]).

.

Part III Option-Theory Loan Risk Model Shortfall risk on uncovered loans to companies on the basis of an option-theory approach Loans covered against shortfall risk Calculation of the combination of loans with the lowest interest costs .

.

More recently this approach has been taken up again by Grenadier [GREN96]. A company will be individually assessed. Here an inference was made from the volatility of the stockmarket prices of listed companies to the volatility of their values. We therefore introduce a fundamentally different way forward in this chapter. for which the appropriate facts and contexts have to be known. TOGETHER WITH FURTHER ELABORATION It was Black and Scholes who ﬁrst described the equity of a company partly ﬁnanced by outside capital as a call-option and the debts as risk-free credit. combined with a put-option on the total value of the company [BLSC73. with debts no longer being examined in the form of a zero bond. Loans to non-listed companies may thereby be assessed also. to assess the volatility of values of the company. but pursues. the KMV Corporation in San Francisco. It will rather be assumed here that the repayment of the loan paid out L plus the agreed interest at the rate of i is owed at the end of any period of time. however. S. The Black/Scholes approach is developed further here. Cox and Rubinstein [CORU85.2.2. when it comes to the statistical determination of the shortfall risk of uncovered loans to companies. as the nominal value of a zero bond may also be seen as its market . California. starting likewise from a stochastic process. a similar solution approach. 375 and thereafter] described this approached later. use the Black/Scholes formula. is adopting a similar approach (see [VAS184] and [KAEL98]. in the context of this study. and the Black/Scholes formula will be used as a model. This cannot be satisfactory. In contrast to this. As already mentioned in Section 1. 637 and thereafter]. As a second step it will be demonstrated. 7. The individual assessment of any one company is only possible by cross-comparison. In this their reﬂections were based on the assumption that the outside capital was made available in the form of a zero bond. The relevance and necessary scope of the facts is uncertain. Their model does not. This is completely permissible. The largely inadmissible extrapolation from the past into the future can only be avoided by using economic forecasts.1 DIFFERENCE IN APPROACH BETWEEN BLACK/SCHOLES AND KMV. on the basis of operational information. how loans to private individuals may be assessed too.7 Shortfall Risk on Uncovered Loans to Companies on the Basis of an Option-Theory Approach There are some hurdles to be cleared. as has been shown in Section 6. S. with further reﬁnements. in so far as the facts needed may exist. USA. on the strength of analogous conclusions. the volatility of the capitalised free cash ﬂows will be used. using an option-theory approach.

1 and 5.11. a yield to be achieved that determined the relationship between market value and nominal value. This is permissible.2 DERIVATION OF BASIC FORMULAE First of all debts are assumed to be in the form of one single bank loan. In applying these considerations to loans. in the next section. as the basis for an investment in zero bonds. application of the method will be illustrated with the aid of examples. It becomes possible.12. obtained by applying equations (4. L · (1 + i)) V =E+D E = equity D = debts V = value of the company At the end of the period the market value of the company’s equity corresponds either to the company’s market value V less the bank’s demand L · (1 + i) — or it is. 7. The result has to be the same in the end. and the most important results will be portrayed in Section 7.7 and 7. in which at the end of a period the nominal value L plus the agreed interest at the rate of i will become due for payment according to the period concerned. if the rate of interest i corresponds to the length of the period. this approach leads to a solution for the risk-adjusted loan interest rate which.3) . As will be demonstrated in this chapter. in the case of the (7.1.8. The intermediate assessment of outstanding credit will be discussed in Sections 7. is independent of the risk-free rate of interest i s . in a simple way.29) respectively. however. In Chapter 8 we will see that it is also possible to assess companies that have taken up both covered and uncovered loans at the same time. in contrast to that of Black/Scholes. following on [CORU85. 0) D = Min(V . 377]: E = Max(V − L · (1 + i).1) (7. 7. corresponds to the return on the market value set by the investor over the remainder of the term. and the inﬂuence of the privileged wages and salaries payable in the event of bankruptcy.60 Risk-adjusted Lending Conditions value with interest on it accrued. will be examined in Section 7. On the basis of these assumptions the following possible values result at the end of the period for the equity and for the debt. The nominal value X of the zero bonds according to the notation by Black/Scholes may thus be seen as shorthand for X = L · (1 + i).3 to 7.2) (7. The derivation and discussion of the risk-adjusted loan interest rate at the time of establishing the loan conditions will follow in Sections 7. S. Λ = L will be put in again. as we did in Sections 4.9. The length of the period of time will ﬁrst of all be consciously left open. under Switzerland’s laws concerning the pursuit of debt and bankruptcy. Every investor is taking. therefore — unlike the approach of Black/Scholes — to assess debt structures that may be as complicated as you like.6. Section 7. in which Λ corresponds to the current market value and i. This consideration leads logically to the conclusion that the nominal value of the zero bond may be replaced by the magnitude Λ · (1 + i). The ﬁnal section. The fundamental of the methods being developed here will be explained ﬁrst.49) and (5. A generalised solution is.10 will specify the limits on the application of the method presented here.

S.2 bank’s demand being higher than the company’s market value. as it is not worthwhile for them to pay back a loan demand that is higher than the company’s market value.1 E = V − L(1 + i ) V D L(1 + i ) D = L(1 + i ) D=V L(1 + i ) V Figure 7.2 show the situation at the end of the loan’s term. or the owners of the company let it go bankrupt. The call put parity leads to. equal to zero. On the other hand.4. Figures 7. 380]. This is expressed in equation (7.3 and 7.Shortfall Risk on Uncovered Loans: Option-Theory Approach 61 E E=0 L(1 + i ) Figure 7. In this situation it is more advantageous for the owners of the company to invest the money in a new enterprise. On the basis of the ﬁgures it becomes furthermore clear that L · (1 + i) corresponds to the striking price and V to the basic value of the option. The unexpired term of these options corresponds in each case to the time until the loan falls due [CORU85.3). From Figure 7.1 and 7. at the end of the term of the loan the bank receives either its demand L · (1 + i) paid back. with P corresponding with the value of the put: E = P + V − L · (1 + i) · (1 + i s )−1 (7. 7.2. the debts may also be portrayed as a risk-free investment with the value L · (1 + i) plus a European put option written on the company’s market value. On the other hand.1). Debts and the market value of equity together result in the company’s value according to equation (7. according to Figures 7.1 it becomes evident that the market value of equity is no different from the equivalent of a European call option on the company’s market value.4) .

as follows: D=V−E=L· 1+i 1 + is −P (7. The quotient of the expectation value and of the fully discounted nominal value of the loan demand is here no different from the survival chance of the loan taking into account breakdown distributions. likewise matching the period of time.3 D D = L(1 + i ) D=V L(1 + i ) Figure 7.5) The value of the debts works out. using equation (7.7) . This value may now be compared with the value of the debts.62 Risk-adjusted Lending Conditions L(1 + i ) L(1 + i ) V − L(1 + i ) −L(1 + i ) V Figure 7.3). It works out for the company’s value V as follows: V =L· 1+i 1 + is +E−P (7.6) The expression L · (1 + i) · (1 + i s )−1 is no different from the fully discounted nominal value of the loan plus interest.4 V The value for the risk-free standard rate of interest i s must of course be put in. χ∗ = D L· 1+i 1 + is = 1 − ρ∗ (7.

14) may be remodelled.12) L· L 1+i − 1 + is 1+i · 1 + is P (7. in the case of only one bank loan.14) (7.11) (7. the quotient in the brackets of equation (7.10) (7.8) − ρ∗ · L · ρ∗ · L · ρ∗ = 1+i 1 + is 1+i 1 + is P L· 1+i 1 + is =L· =P 1+i 1 + is Equations (7. which demonstrates the relation between the risk-free standard rate of interest i s .3 DERIVATION OF RISK-ADJUSTED VALUES First the assumption is made of a company that has its debts position only in the form of a bank loan. 7.11) and (7.9) (7.15) The result reads: 1+i 1 = 1 + is 1 − ρ∗ (7.13) Now comes the decisive difference from the approaches to date made by Black/Scholes and Merton.49).12) give in detail the correlation between the put value and the credit shortfall risk probability taking breakdown distributions into account. then the value of the debts corresponds.7) results in: 1 − ρ∗ = Further transformations lead to: (1 − ρ ∗ ) · L · L· 1+i 1 + is 1+i 1 + is =L· 1+i 1 + is −P −P (7.13) one obtains: L=L· 1+i 1 + is −P (7. If it is intended that the loan interest rate i be risk adjusted.Shortfall Risk on Uncovered Loans: Option-Theory Approach 63 Equation (7.16) .6) substituted into equation (7. With the aid of equation (4.6) into equation (7. to its nominal value: D=L By substituting equation (7. s +ρ 1 + i1 − ρ ∗ 1 + is 1 − ρ ∗ + is + ρ ∗ 1+i = = = ∗) 1 + is 1 + is (1 + i s ) · (1 − ρ (1 + i s ) · (1 − ρ ∗ ) ∗ (7. the shortfall risk ρ ∗ and the risk-adjusted rate of interest i.

26) (7. lead in what follows to the general solution. If χ ∗ = 1 − ρ ∗ (see equation (2.17) Elimination from equation (7. already mentioned in Chapter 1. which does.18) it follows from P = 0 that ρ ∗ = 0.25) (7.27) P L+P (7.23) P · (1 − ρ ∗ ) L (7. and vice versa.24) . the credit shortfall risk starting from equation (7.20) P L· 1 1 − ρ∗ (7. detached from the risk-free standard rate of interest i s and from the loan interest rate i.23) it is the case that: 1 − χ∗ = Transformations result in: L + P − χ ∗ (L + P) = P χ ∗ · (L + P) = L L χ∗ = L+P (7. if one succeeds in calculating the value of the put. the later calculations can be simpliﬁed. Under equations (7.21) (7.16) expressed as follows: ρ∗ = Extended by (1 − ρ ∗ ) results in: ρ∗ = Further transformations result in: ρ∗ · L = P − ρ∗ · P ρ ∗ · (L + P) = P P ρ∗ = L+P (7. and for P = ∞ that ρ ∗ = 1.8)) is calculated ﬁrst. To test this.23) are identical.18) From this it follows that the credit shortfall risk for the calculation of the risk-adjusted loan interest rate is known.18) and (7. Equation (7. What is essentially new in the approach presented here is that the relation between i s . All values that P may assume thus result in a permissible value for ρ ∗ . This is the small detour.14) this gives: L= L −P (1 − ρ ∗ ) P L+P (7.12) is. however. According to equation (7. with the help of equation (7.12) will be calculated once more.64 Risk-adjusted Lending Conditions Substituted into equation (7.16) is used.17) using ρ ∗ leads to: ρ∗ = (7. ﬁrst to calculate the credit shortfall risk ρ ∗ .18) and (7. i and ρ ∗ according to equation (7.22) (7.19) The comparison shows that equations (7.

32) (7.16) and (2. This is the equivalent of granting credit to a borrower that would be. If the rate of interest is in fact established as risk adjusted. as the following consideration indicates: the alternative investment is made again at the risk-adjusted rate of interest to a borrower that is real and does exist.30) d+ d χ∗ d · N (x) − N (x − σ ) (7.33) (7. that a bank’s yield in any loan transaction always corresponds to the risk-free standard rate of interest i s . however.27) expresses itself. the value of P amounts to: P=L· (1 + i) · N (x) − V · N (x − σ ) (1 + i s ) ln L · (1 + i) V · (1 + i s ) (7. in other words. providing the loan interest rate i is in fact calculated so that it is risk adjusted. As such borrowers do not in reality exist this assumption might appear fanciful. With the deﬁnition of debt rate d= d = debt rate equation (7. the bank’s yield on this investment is.Shortfall Risk on Uncovered Loans: Option-Theory Approach 65 According to Black/Scholes.34) (7. for a period of time following on [CORU85.35) (7.28). after deduction of losses. 211]. The i s may thus take over the function of the risk-free rate of interest in terms of being a model in the context of the lending operations of any bank. again equivalent to the rate of interest i s .31) (7. risk-free. This is not.29) with: x = N = standard normal distribution function σ = volatility of the company value according to the time period On this we make the assumption here that the alternative investment might be made at the i s rate of interest. (7. as follows: χ∗ = Multiplying out results in: χ ∗ · (d − N (x − σ )) + d · N (x) = d Reverse substitution χ ∗ = 1 − ρ ∗ results in: (1 − ρ ∗ ) · (d − N (x − σ )) = d · (1 − N (x)) ρ · (d − N (x − σ )) = d − N (x − σ ) − d + d · N (x) d · N (x) − N (x − σ ) ρ∗ = d − N (x − σ ) with: x = ln d 1 − ρ∗ ∗ σ + σ 2 L V (7. S.28) (7. the case. by applying equations (7. This means. for the bank.36) σ + σ 2 .8) abbreviated to V .

the duration of these also has matching inﬂuence.37) ρ (t) = √ N (x − σ · t) − d √ ln 1 − d ∗ (t) σ· t ρ (7.66 Risk-adjusted Lending Conditions As ρ ∗ also occurs in the expression for x. This value has to be converted to the mean credit shortfall risk per annum. 382]. see Appendix 2). Volatility relating to one year will be assumed for σ .38). If various new loans with different terms are granted to a company. In the reverse case. and/or the remaining term of a loan that it is intended to reassess.2). Highly interesting is the fact that the credit shortfall risk ρ ∗ is now only dependent on the degree of outside indebtedness d and on the volatility σ of the company’s market value.37–7.39) The annual rate of interest consistent with risk can then be calculated with the aid of equation (4. In particular the credit shortfall risk is. may be chosen as the period of time. the set of equations (7.35)–(7. It follows from this that ρ ∗ = 1. This means that ρ ∗ approaches 1 for high (1 − ρ ∗ ) degrees of outside indebtedness. As the volatility is. the credit shortfall risk is also approaching zero. The solution thus runs as follows: √ N (x − σ · t) − d · N (x) ∗ (7. aided by the rules for transforming notice periods (Section 3. S. 205] N (x) ≈ N · (x − s) ≈ 1 applies where d/ 1. where d/(1 − ρ ∗ ) 1. The approximation [CORU85.36) must be solved iteratively.49): i pa ∗ i spa + ρpa ∗ 1 − ρpa (7. This means if the outside indebtedness approaches zero.40) i pa = interest rate on annual basis i spa = risk-free interest rate on annual basis ∗ ρpa = average annual credit shortfall risk The annual volatility should be inserted here for σ . dependent on the time period and/or loan term under consideration. not dependent on the risk-free standard rate of interest i s ! This is the essential difference from the Black/Scholes result using the zero bond approach [CORU85. the approximation N (x) ≈ N · (x − s) ≈ 0 applies. then the appropriate credit shortfall risk must be decided for each term and proceeded with as described above. as it is annual accounts and budgets that are normally evaluated in the context of loan assessment. It follows from this that ρ ∗ ≈ 0. and the term in years for t. however. The term of the loan that has been newly paid out. ∗ ρpa = 1 − t 1 − ρ ∗ (t) (7. This presents no problem with today’s standard PC software (for example Microsoft Excel Release 4 or higher. S. in order to be able to calculate the annual rate of interest.38) + with: x = √ 2 σ· t t = loan/term It is important to note here that ρ ∗ (t) is calculated with equations (7. however. ρ ∗ (t) is thus the probability of credit shortfall over the whole loan term. .

means the following: r Either. If there is no such hope. it may in all probability be assumed that this ﬁrst assumption applies in the case of ﬁrms of good ﬁnancial standing. Owing to his inability to make interest payments. that loan interest is only paid for the whole term of the loan when the loan is repayable. the time between two assessment dates should be inserted for t. the borrower may thus already go bankrupt. The analogous conclusion of how to proceed for a private borrower is put forward in a ﬁnal subsection. this makes complete sense. Compare this to Brealey and Meyers [BRMY96. it does not have to be waited for. This is of course not always the case! What does this now imply for the application of the method described here? Even if interest payments become payable during the term of a loan. however.4 DETERMINATION OF THE VALUES FOR THE SOLUTION FORMULA This chapter is concerned with giving the user of our method useful hints on how it may be applied in practice. In this case the market value of the assets corresponds to the market value of the equity. then it makes no difference if the borrower goes bankrupt earlier or only when repayment of the loan becomes due. as the bank’s time for reaction becomes shorter. always apply in the case of ﬁrms of poor ﬁnancial standing. possibly until the Only empirical investigations may show whether or not these implicit assumptions are permissible. S. The second assumption will. As has been shown. before the loan itself is in default. as the owners of the equity do not have to fulﬁl any expectations for providers of outside capital.4. First we explain how to proceed when the borrower is a company. that the submission of intermediate accounts has an improving effect on credit-worthiness. how interest payments and loan repayments may be taken into account prior to expiry of the loan term. r Or. and a correspondingly shorter period may be inserted into the formula for t. that the probability that the borrower already goes bankrupt owing to a loan interest payment becoming due. At this point the supposition may merely be expressed that they do apply. The debt-free company serves as a starting point for our considerations. It lines up with the current banking practice of demanding intermediate accounts of borrowers at higher risk.Shortfall Risk on Uncovered Loans: Option-Theory Approach 67 In the case of current account loans. it may happen that the borrower is not in a position to pay them. For this it is assumed that the borrower’s track record and budget documentation are available to the user. i. 668]. on the basis of the approach used here. This is not taken into account here. normally one year. 7.3 we demonstrate. however. It is still being assumed. by implication. That. that the borrower has been given time to make the interest payments. . If the put remains in the money in any case. if the bank extending the credit is justiﬁed in hoping for a recovery in the situation.1 The Value of the Company and its Debt Rate The expositions in this subsection are based on explanations by Brealey and Meyers [BRMY96]. with the help of a real-life example. It follows from this. 7. In Section 11. repayment of the loan is due. is so small that it may be ignored.e.

which comes into place for the method described here. It would.41) is calculated as follows [BRMY96. The discount rate used according to equation (7. This may mean. The free cash ﬂow is here deﬁned as revenue less costs less investments [BRMY96. Sufﬁce is to point out here that in today’s literature.42) (but also compare equation (7. is therefore calculated as follows: id · L L = (7. S.44) d= V C Summarising.44): the debt rate relevant to lending is independent of the debt rate used for accounting.41) The total according to equation (7. S. in an extreme case. 49]: V = C id (7. S. the value of a company may be determined by discounting its free cash ﬂows and totalling the resulting values [BRMY96. along with free cash ﬂow.46) and the statements made there).68 Risk-adjusted Lending Conditions According to Brealey and Meyers. 143–236].42) is. S. despite some shortcomings that are widely recognised. of the average. however. 71]: Free cash ﬂow = Revenues − Costs before interest and taxation − Investments It is important to emphasise at this point that the free cash ﬂow is calculated with costs included prior to interest and taxation. go far beyond the scope of this study to go any more closely into the determination of the discount rate at this juncture. 72]: V = C = free cash ﬂow i d = discount rate ∞ n=1 C (1 + i d )n (7. for example [BRMY96. future free cash ﬂow of the company. the value of companies is therefore calculated as follows [BRMY96. the CAPM (capital asset pricing model) is as favoured as ever: i d = i g + β(i mt − i g ) β = measurement of unleveraged assets market risk according to CAPM i g = return on a risk-free investment in government bonds i mt = return on the market We refer you to the literature concerned for further expositions on the CAPM. An importance consequence ensues from equation (7. S. without bring the liabilities side of the company’s balance sheet into the picture at all! In the notation being used in this study. let it be emphasised that it is a question. in the case of C. We are indeed concerned here at ﬁrst to determine the value of the assets alone. that a company that is today overloaded with debt (7. The debt rate. of decisive importance.43) . 71/71]. Only the discounted future free cash ﬂows count.

S. only empirical enquiries will show. Γ = gamma function n = number of quotients of annual reports and budgets . however.Shortfall Risk on Uncovered Loans: Option-Theory Approach 69 according to its books may still be credit-worthy. overloaded with debt. Substantially better results in determining volatility could be achieved if borrowers produced accounts every six months or even quarterly. The correction factor in front of the root in equation (7. 7. according to its books. The method being described does therefore provide the answer in relation to the extent to which it may be worth restructuring a company that is. that the structure of the company’s ownership does. 254 and thereafter]: σ = Γ n−1 2 Γ n 2 · 1 · 2 1 · n n ln k=1 Vk+1 Vk 2 −µ (7.4. however. as considerable work would be involved and no one apart from banks would be asking for them. 159]. Better to combine the two. even though these are associated with uncertainties. using annual accounts going back three or four years and one to two future-orientated budgets.2. On a purely theoretical basis alone. it is the case that the future is no simple extrapolation of the past. when all is said and done. provided its future prospects are otherwise good. substantially longer series of facts would be available for recently observed periods of time and results would improve accordingly. The concrete calculation of annual volatility takes place by analogy with the method that was described by Cox and Rubinstein [CORU85.45) for various values of n is shown in Table 7. This can only be determined with the help of empirical investigations (see Chapter 10). Whether the very simple procedure we sketch out is sufﬁcient to do justice to the method described here.2 Volatility In order to gain some idea of the volatility of a company it may.1 [KREY91. If such intermediate accounts did exist. But this is often not the case. In so far as intermediate accounts do exist.45) n µ= ln k=1 Vk+1 Vk µ = medium of the logarithms The gamma function values are of interest as shown in Table 7. it is therefore recommended that they be used. on the one hand. it is not possible to lay down how volatility should be determined correctly for the purposes of the model. Here again. As long as there are banks that do not insist on intermediate accounts this is not going to gain acceptance in the market. be ascertained on the basis of historical data. S. Recently Cho [CHO98] has shown. have inﬂuence on the company’s value. This study shows once more how complex is the question of valuing companies. with the aid of empirical investigations.

[HULL97. S. From the point of view of the lending bank. It has only been possible here to show a practical method that is effective in application. In relation to earlier annual accounts. The difﬁculties in determining volatility can be seen in.5 3· √ 4 π 3 2 2.2 Correction factor values n Γ n−1 2 Γ(n ) 2 2 1.8862 1. which must be substituted into equation (7.46) for V should also be used in equation (7.7725 0.7725 3 1.6).0000 4.0000 2. The only difference is the deﬁnition of σ 2 as the average over the remaining term T of the option: σ 2 (T ) = T −1 · 0 T σ 2 (t) dt (7. Determination of the relevant volatility plays a central part in the whole option theory.3293 3. for example [SWD192].0000 3.7523 6 0. especially in cases of winding up and reconstruction.6018 8 0. Merton [MERT73] showed later that the Black/Scholes model may be applied as it were unchanged.000 1.42).0000 1.44).5 √ 105 · π 16 5 24 24. equation (7. strictly speaking. The credit-worthiness and ﬁnancial standing of any company in fact depend. The expositions given in the previous subsection apply regarding this.6647 7 0. to have an adequately accurate idea of σ 2 over the loan’s term.4847 The values of a company for individual ﬁnancial years (historical annual accounts and budgets) should likewise be calculated according to equation (7. The reﬂection that every company disposes of a liquidation value. This leads to the following equation for the value of a company. more than anything on liquidation value (see also Section 7. [BCCH97] and [RITR98].1 Gamma function values x Γ (x) 0. to be able to apply the method described here.70 Risk-adjusted Lending Conditions Table 7.47) It is thus sufﬁcient.631 Table 7. 499 and thereafter]. if volatility is not constant but is known as a function of time. The Black/Scholes model assumes volatility is constant.42): Vk = Max Ck .46) If necessary. we must — at this point — go particularly into years with negative free cash ﬂows: it makes no sense to discount negative free cash ﬂows.5 √ 15 · π 8 4 6 6.8862 5 0.5539 9 0.3234 11. [KAR093]. [RSS093].1284 4 0.5 √ π 2 2 1 1. the value of a company is never lower than the liquidation value that can be expected.5158 10 0.5 √ π 1 1 1. even if it may only be very small. is of further assistance here. Vlk i dk (7. .

by talking to each individual borrower. and these will be described in Chapters 8 and 9. For this reason. The CAPM may not be applied here. this is normally the legal interest rate cap. The more luxuries come to be regarded as necessities the smaller the borrowing power of the private individual becomes. but the following reﬂection helps further. the rules for covered loans apply. So in granting credit to individuals it is extremely important to take out appropriate insurance and to ensure the proceeds of it are assigned in favour of the bank. is of decisive importance. It is therefore appropriate to apply the highest possible loan interest rate that could be expected. Only discretionary income has been taken into account in the reﬂections outlined here — any assets that may exist have not been considered. the credit shortfall risk according to our model depends on the following inﬂuence factors: r The company’s debt rate d r The annual volatility of the company’s value σ r The loan’s term in years t .3 Private Debtors The situation as regards private debtors can be shown by analogy with that of companies. First the rate of unemployment beneﬁt applicable.5 INFLUENCE OF INDIVIDUAL PARAMETERS ON THE CREDIT SHORTFALL RISK As was shown in the previous section. In the case of consumer credit. with which the quasi-market value of the individual is found out. The risk that the borrower may become unemployed must be taken into account by an adjustment to the ﬁgure for discretionary income. With private debtors free cash ﬂow is the equivalent of discretionary income after deduction of essential living costs. occupation and residence. As no asset should be taken into account. and vice versa. namely that the capability. Next we have to determine the appropriate discount rate i d . such that one obtains the expectation value of the discretionary income taking the risk of unemployment into account. must be determined. In this way loans to individuals also become assessable. The former risk may be covered by the insurance industry on attractive terms. taking into consideration any possible repayments. for the purposes of discounting.4. then the business does not get done anyway. By contrast with companies. If the method outlined here ends up with a loan interest rate that is higher than the legal interest rate cap. to make interest payments and repayments. Then the ﬁgure for discretionary income must be multiplied by a factor of 1 less the above-determined unemployment beneﬁt. is a tricky task for bankers. 7. by way of precaution. two further risks have to be assessed in connection with an individual’s discretionary income: the risks of death (and incapacity to work) and of unemployment. Once discretionary income for a few years back has been determined in this way — and used as a budget for the future — these values may be used instead of the free cash ﬂows in the formulae described above.Shortfall Risk on Uncovered Loans: Option-Theory Approach 71 7. on the basis of the borrower’s personal data in respect of age. unless it is assigned in favour of the loan. in the case of the individual. determining the correct ﬁgure to put on discretionary income at any one time.

Figure 7.5. providing the volatility of the company’s value is correspondingly low. Figure 7. Even though it has not been possible to date to carry out any costly empirical tests — for the reason that no bank in Switzerland has the necessary historically documented data available — the following diagrams do in fact line up qualitatively with the author’s professional experience to date. but the credit shortfall risk axis has been spread out. then the risk exposure is 100%! Figure 7. which goes back at least 15 years. even in the case of very high debt rates such as those of banks and insurance companies.72 Risk-adjusted Lending Conditions Parameter: 100% 50% 0. .5 0.7 corresponds with Figure 7.5 0.5 shows that an appropriate credit shortfall risk switches in according to the volatility of the company’s value. Here it is evident that low credit shortfall risks result.6 corresponds with Figure 7. that the following always applies: ρ ∗ (d = 100%) = 100%! If the loan granted is higher than the company’s market value. with the help of the following diagrams. Here it is evident that low credit shortfall risks result. once a certain debt rate has been passed.6 to the extent that the debt rate axis has been spread out too. It is important to note. at this point. even in the case of high debt rates. provided the volatility of the company’s value is very low.2 d 100% It is intended to demonstrate the inﬂuence of these parameters on the credit shortfall risk according to our model.3 0% 0% 50% t=1 ρ*(d = 100%) = 100%!! Figure 7.

1 0.6 Parameter: 2% 0.5 0.7 .05 0.02 1% 0.1 0.05 0% 0% 50% t =1 100% d Figure 7.01 0% 90% 95% t=1 100% d Figure 7.2 1% 0.Shortfall Risk on Uncovered Loans: Option-Theory Approach Parameter: 2% 73 0.

74 Risk-adjusted Lending Conditions Parameter: d 100% 80% 60% 50% 40% 20% 0% 0 0. but over shorter loan terms. The dependence of the credit-worthiness of a loan on the term of the loan with the same borrower. Figure 7.5 t =1 1 Figure 7.8 suggests the conclusion that more than anything else it is important for an entrepreneur to keep the volatility of the market value of his company low. Comparison between Figures 7. Figure 7. and not necessarily its debt rate (compare comments in Section 9. According to the guidelines of the rating agencies. provided the debt rate is correspondingly low. Figure 7. but over shorter loan terms.11 corresponds to Figure 7. Here it is evident that low credit shortfall risks result. if the shortfall risk amounts to less than one millionth over the whole term.8 corresponds to Figure 7. Figure 7.10.12.8.10 demonstrates the course of the credit shortfall risk independently of the term of the loan and of the volatility of the company’s value.12 corresponds to Figure 7. Figure 7.4). with the credit shortfall risk axis again spread out.9 corresponds with Figure 7.14 shows which pairs of values (σ . but σ and d have been exchanged. d) match the same credit shortfall risk.10. even in the case of high volatility in company value. Here too it is evident that low credit shortfall risks result even in the case of high debt rates.5. provided the volatility of the value of the company is low.13 corresponds to Figure 7. Figure 7. The line with ρ ∗ (t) = 1 ppm therefore represents the AAA curve to the extent that all pairs of values below this curve match this top rating. where the volatility of the company’s value and its debt rate have changed places.8 Figure 7. . is clearly discernible.5 and 7. a debtor receives a AAA/Aaa rating.

9 Parameter: 5% 4% 0.1 0% 0 1 2 3 4 t 5 Years d = 60% Figure 7.Shortfall Risk on Uncovered Loans: Option-Theory Approach Parameter: d 2% 80% 60% 75 1% 40% 0% 0 0.4 Figure 7.2 3% 2% 1% 0.10 .2 t=1 0.5 0.

2 0% 0 3 6 d = 60% 9 12 t Months Figure 7.35 1% 0.5 0.76 Risk-adjusted Lending Conditions Parameter: 2% 0.11 Parameter: d 5% 80% 4% 60% 3% 2% 1% 40% 0% 0 1 2 = 0.12 .2 3 4 t 5 Years Figure 7.

14 25% 50% 75% 100% .2 9 12 t Months Figure 7.Shortfall Risk on Uncovered Loans: Option-Theory Approach 77 Parameter: d 90% 5% 4% 3% 2% 80% 1% 70% 0% 0 3 6 = 0.13 d 100% Parameter: *(t ) 75% 50% 1% 1 bp 25% 1ppm 0% 0% Figure 7.

S. n. we can thus write. p] [CORU85. In other words.2. This probability can be determined according to Cox and Rubinstein. Let us remember. This probability receives the value N(x − σ · t) [CORU85.4) by b =1− ρ∗ ρ∗ =1− ρ N (x) ρ∗ ρ (7. As may be inferred from Figure 7.38) σ· √ t √ σ· t + 2 With the credit shortfall risk ρ ∗ determined. too. This leads to the following reﬂections: r Companies of good ﬁnancial standing have high revenues and thus high market values. according to equation (7. the breakdown distribution rate relates to the nominal value of the loan plus all interest and compound interest over the whole term of the loan. at this point. S.78 Risk-adjusted Lending Conditions 7.3. the probability of bankruptcy ρ can thus also be determined. 177].3 in Section 7. that it is the market value of the debt-free company that is . analogously. The probability of the corresponding put at maturity √ being in the money √ is therefore 1 − N (x − σ · t) [CORU85. with the notation introduced there [CORU85] being used in the paragraph printed in italics below. as follows: ρ = N (x) x= ln d 1 − ρ∗ (7.6 RISK OF BANKRUPTCY AND BREAKDOWN DISTRIBUTION Up till now it has been the credit shortfall risk ρ ∗ that has been calculated. in order to be able to understand the derivation better. which are always higher than the balance sheet ﬁgures produced by their accountants (and let us remember at this point. Returning to the notation being used here. that the minimum market value of a company corresponds. At this juncture we should demonstrate how the risk of a borrower’s bankruptcy and the probably breakdown distributions may be calculated.49) and the expected breakdown distributions by: B = b · L · (1 + i(t)) = 1 − ρ∗ N (x) · L · (1 + i(t)) = 1 − · L · (1 + i(t)) (7.50) i(t) = loan interest rate according to the whole term As was emphasised at the end of Section 7. This becomes N(x + σ · t) on the basis of the symmetry properties of the standard normal distribution function [CORU85. according to equation (2. S.48) as per (7. 4].46) to its liquidation value. the calculations in this chapter are based on the assumption that the loan interest for the whole term of the loan is due for payment only when the loan itself is due to be repaid. 208] in the case of transition to the standard normal distribution function. by way of qualiﬁcation. S. The breakdown distribution rate probability is calculated. which should be taken into account here. In the same way the calculation of the breakdown distribution rate is also based on this assumption. by using the√ complementary binomial distribution function. The probability that a call is in the money at maturity amounts to Φ[a. 211]. the probability that a borrower is bankrupt at the time of the loan becoming due for repayment is none other than that of the put being in the money.

the current standard rate of interest appropriate to the remainder of the term must be used for i spa .53) In this. The basic equations from Section 7.52) √ t √ σ· t + 2 (7. where in the risk-adjusted case for d > 1 there is no longer any solution for ρ ∗ (or where ρ ∗ = 100%) (cf. As will be evident from equations (7. in which (1 + i)/(1 + i s ) = 1/(1 − ρ ∗ ) no longer applies. This leads to the following equation. distinguish between the market and the nominal value of the loan. this results in: Λ=L· 1 + i pa 1 + i spa with: x = t · (1 − N (x)) + V · N (x − σ · ln d · σ· 1 + i pa t 1 + i spa √ t) (7. if i pa according to ρ ∗ is consistent with risk! Unlike Section 7. This may lead consequentially to an existing loan in this situation being called in.12). however.4.2.6) and (7. the market value of the company through to the end of the term may improve again. 7.Shortfall Risk on Uncovered Loans: Option-Theory Approach 79 r used for risk calculation). Expressed in .48)– (7. t means the rest of the loan’s term and the interest rates i pa and i spa and the volatility σ must be inserted on an annual basis. according to subsection 7.3 to 7. That leads to very low shortfall risks — in the main just theoretical. as the interest rate during the contractually agreed term of the loan may no longer be adjusted to be consistent with risk: Λ=L· 1 + i pa 1 + i spa t −L· 1 + i pa 1 + i spa t · N (x) + V · N (x − σ · √ t) (7. Even if the value of d is greater than 1 during the term. in which we must. This is explained by the fact that it is indeed normal that conditions may no longer be adjusted to be consistent with risk in new circumstances during the agreed ﬁxed term of the loan. On the other hand companies of poor ﬁnancial standing have low revenues and thus low market values. although new information for assessing it is available. for the risk calculation.6. and the situation of d being less than 1 may be achieved again by the time the loan is due for repayment.3. Λ = L. or to an application for a new loan not being followed up. in which on the basis of the loan agreement the loan interest rate up to the time the loan is due to be repaid can no longer be adjusted to be consistent with risk. which lines up with banking practice.49).2 again form the starting point for our considerations. Moreover. notes on Figure 7. it may thus be assumed that it is the liquidation value that is used. this section will deal with assessing a loan that has already been granted during its term.51) Summarised.7 LOAN ASSESSMENT The derivation of the risk-adjusted loan interest rate when extending credit having been analysed in Sections 7. The value of a loan during its term can be calculated with the aid of equations (7. This gives expression to the fact that it is only the situation when the loan is due for repayment that is decisive. As soon as a serious crisis looms. Loans are thus small in size in relation to the company’s market value.5). there is a solution here for all values of d.

however.52)–(7. in the case of several loans.4). may be calculated on the basis of the stock exchange price.37)–(7.7. Tackled this way we can be certain that the loan in question is assessed according to its characteristics. may be determined using equations (7. ρ∗ ρ · L · (1 + i(t)) − S (7. that does not necessarily mean that it is also in the money at maturity (compare also with Brealey and Myers [BRMY96. and therefore the bond’s rating.50) the breakdown distribution probability value comes to: B = b · L · (1 + i(t)) = 1 − ρ∗ ρ · L · (1 + i(t)) (7. under Swiss debt recovery and bankruptcy law. For the case in which the liabilities side is made up of several loans.8 BONDS Bonds can be assessed in exactly the same way as bank loans. according to Section 7. With the help of equations (7. These salary claims have thus not been taken into account in the considerations in this chapter so far.38).53). the shortfall risk implicit in the stock exchange. The value of the loan being considered then works out in proportion to its share of the debts. With the aid of volatility investigated in this way. the company does not have to account for these salary claims prior to bankruptcy. In contrast to all other relevant liabilities in bankruptcy. Under equation (7. The privileged demand of each person employed is that person’s salary claim up until the expiry of the notice period according to his/her contract of employment. S. Here the values of the loan under consideration must be used for the rates of interest i pa and i spa . the debts as a whole must be inserted for L and assessed according to the remaining term of the loan in question.54) . This implies the following: in bankruptcy the whole amount for salaries is deducted from the overall breakdown distribution. namely the volatility implicit in the stock exchange.80 Risk-adjusted Lending Conditions terms of option theory: even if the put during the loan’s term is occasionally in the money. 7.50) First of all the privileged salary claims have now to be deducted from the expectation value of the breakdown distribution available for distribution: Bc = b · L · (1 + i(t)) − S = 1 − Bc = corrected breakdown distribution S = proportional salaries It should be noted that. the converse. but taking into account the total of debt (see also Section 8. This shall now be remedied. 564]). only the proportional salaries have to be taken into account for any individual loan that has to be assessed. 7. Under this the date on which bankruptcy proceedings are commenced is taken as the date on which notice of termination of employment is given.9 CONSIDERATION OF PRIVILEGED SALARY CLAIMS IN THE EVENT OF BANKRUPTCY The salary claims of a company’s workforce are privileged in the case of bankruptcy.

6): ∗ ρc = ρ · (1 − bc ) = ρ · 1 − ∗ ρc = corrected credit risk Max(b · L · (1 + i(t)). the corrected credit risk may be calculated by: i c (t) = ∗ i s + ρc i s + ρ · (1 − bc ) = ∗ 1 − ρc 1 − ρ · (1 − bc ) (7.50) must therefore be correctly written as follows: Bc = Max(b · L · (1 + i(t)) − S.59) are used. It is quite possible that the salary claims in bankruptcy are higher than the probable breakdown distributions.61) . this means nothing more than that it has to make a proportionate contribution to the privileged salary claims that have to be met. which would make no sense.59) and (7.58) For other cases the following applies: ∗ ρc = ρ · (1 − bc ) = ρ · 1 − Bc L · (1 + i c (t)) (7.Shortfall Risk on Uncovered Loans: Option-Theory Approach 81 The rest is distributed proportionately across the individual loan demands. however. the solution is: ∗ ρc = ρ if bc = 0 (7. The middle and right-hand term of equation (7. From the point of view of the individual lender. 0) Bc = L · (1 + i c (t)) L · (1 + i c (t)) (7.60) now permit calculation of the corrected breakdown distribution rate probability bc . subtract from 1 and cancel out by −1 results in: bc = Bc Bc = i s (t) + ρ · (1 − bc ) L · (1 + i c (t)) L · 1 + 1 − ρ · (1 − bc ) (7. This just means that the salary claims are no longer covered at all. the corrected breakdown distribution rate probability can be detailed as follows: bc = (7. from the breakdown distributions that were ‘originally’ available to it. The percentage rate of the salary total falling against it in this way corresponds to percentage ﬁgure of its loan claim in relation to the total loan. The corrected credit risk can now be calculated with the aid of equation (2.59) On the other hand. 0) Max(b · L · (1 + i(t)) − S. Equation (7. Thus a negative ﬁgure for corrected probable breakdown distributions would result. 0) L · (1 + i c (t)) (7.55) Continuing.57) For the case where bc = 0.56) bc = corrected breakdown distribution rate i c (t) = corrected loan interest rate according to the whole term It is essential at this point to note the distinction between i(t) and i c (t)! i(t) concerns the uncorrected rate of interest that has already been calculated. Cancel out with ρ. while i c (t) concerns the corrected value still to be calculated.60) Both equations (7.

with the terms having been ‘translated’ for the application described here. The method described here takes the free cash ﬂows of the debt-free imaginary company as the basis for assessing its credit-worthiness. The assessment rests thus on the amount of revenue before interest and dividends. The level of interest is constant.10 LIMITS TO THE APPLICATION OF THE OPTION THEORY APPROACH The preconditions that have to be fulﬁlled in order to be able to apply the Black and Scholes model are speciﬁed in Cox and Rubinstein [CORU85. of a theoretical construction and not of anything existing in real life.39).62) and multiply out gives: bc · L · (1 + i s ) = Bc − ρ · Bc + ρ · Bc · bc Solving by bc leads to: bc = Bc · (1 − ρ) L · (1 + i s (t)) − Bc · ρ Bc · (1 − ρ) L · (1 + i s (t)) − Bc · ρ (7. taxation or transaction expenses. 1.64) (7. The company does not make any distribution payments (dividends and interest) during the period under consideration. As was shown in this chapter. 2. independently of the other ﬁnancial years. in contrast to the solution as it was indicated in Cox and .62) Reduce to the lowest common denominator using the denominator in the right-hand term of equation (7. In the assessment this ﬁgure is determined for each ﬁnancial year being taken into consideration. 7. This precondition is fulﬁlled as it is just a question. as long as the approved and calculated loans are not changed (raised) as a result of the payment of dividends and interest.82 Risk-adjusted Lending Conditions Transformation of the denominator in the right-hand term of equation (7. 4. the credit-worthiness of a company is on the one hand independent of the interest level.65) This credit has still to be converted on an annual basis according to equation (7. There are no requirements in terms of margins. The distributions of dividends and payment of interest thus do not have any inﬂuence on the assessment in the method described here. They are explained below. S.61) gives: bc = L· Bc (1 + i s (t)) (1 − ρ · (1 − bc )) = Bc · (1 − ρ · (1 − bc )) L · (1 + i s (t)) (7. 3. in the case of the put in the application described here.63) The corrected credit shortfall risk is thus: ∗ ρc = ρ · (1 − bc ) = ρ · 1 − (7. Notice to terminate the loan may only be given at the end of the period under consideration. This precondition is always fulﬁlled. 268].

in which the value of the asset base is changing continually but only in small steps. This precondition may usually be regarded as fulﬁlled. that has been obtained as a result of a random sample. but not over longer examination time scales. The precondition of constant volatility was qualiﬁed by Merton [MERT 73]. S. as we have already mentioned: compare also subsection 7. Kremer and Roenfeldt [KRRO92] moreover pose the question of whether the Black and Scholes model is usable at all for longer-term examinations. Translated into extending credit. they conclude that their ‘jump-diffusion’ model in the case of warrants exerciseable in more than one year (for out of money warrants) may deliver more reliable results than the Black and Scholes model.1). use of the visual test by means of a probability paper (for example. Only the calculating out of various scenarios can show whether or not the uncertainties are decisive in terms of price. any substantial events changing the nature of the company. It must not. however. BOHL92. S. Major changes are often attributable to errors in earlier assessments and are therefore artiﬁcial.Shortfall Risk on Uncovered Loans: Option-Theory Approach 83 Rubinstein [CORU85. major changes in commercial strategies and so on. By this we mean events such as mergers. either the price has to be set in line with the worst-case scenario. Even Cox and Rubinstein write [in CORU85. S. One qualiﬁcation results from the point that track records may only be used as far back as there may not have occurred. 7. S.2). The volatility of the company’s value is constant. There are statistical tests to check whether any existing frequency distribution out of n observations (here: appraised company accounts and budgets). BOHL92. and these tests are described in the reading (for example. S.2 with the reading references given there. this means a loan assessment endures only so long as the company laying claim to it is not subject to any substantial changes of the kind mentioned above. The further back a company’s track record goes. Only very small changes in the valuation of a company can occur in very short periods of time. 629]. 5. 258] that ‘the critical feature for option pricing is the behaviour of the volatility’. Even . or no loan must be made at all. 382] (see Section 7. S.4. Changes in the value of a company must be distributed lognormally. the appropriate consequences have to be drawn. 625 and thereafter). In their article about warrants. The requirement that loan agreement clauses must contain appropriate passages follows on from this. Because of the statistically small volumes of data in the form of appraised company accounts and budgets. The approach described here may only be applied if reliable assumptions are made about future volatility. be denied that volatility represents the crux of all applications of option price models. in the period concerned. In so far as this is the case. is consistent with a hypothesis that has been drawn on distribution in the parent population (here: lognormal distribution). the more precisely can it normally be established whether or not the condition is fulﬁlled. The Black/Scholes assumption within this time scale. In any such case the errors should be corrected and the assessment undertaken afresh. The assessment of the market risk on the other hand was intentionally excluded here (see Section 1. 276]. 625 and thereafter) is recommended — it being possible these days to run this up on to a screen with the help of the computer [BOHL92. They argue that ofﬁcially quoted options have a maximum term of 270 days. 6. divestments. may well be permissible. A further precondition is mentioned in Cox and Rubinstein [CORU85.

is of course not the case with recently founded enterprises.52) with: x = √ σ· t + 2 (7. A company must after all have a ‘track record’ in the ﬁrst place.37) (7. to adjust the loan conditions to new circumstances.84 Risk-adjusted Lending Conditions for long-term loans.38) As ρ ∗ (t) in the above equations corresponds to the credit shortfall risk over the whole term of the loan. Assessment is made in the case of t years according to the equations: Λ(t) = L · 1 + i pa 1 + i spa t · (1 − N (x)) + V · N (x − σ · ln d · √ σ t 1 + i pa t 1 + i spa √ t) (7. The only way forward in this situation is to work on the basis of budgets.38). This. with the aid of model calculations and the help of specimen scenarios. 7. to see whether it can be circumscribed with sufﬁcient precision. independently of the risk-free rate of interest: √ N (x − σ · t) − d · N (x) ρ (t) = √ N (x − σ · t) − d √ ln 1 − d ∗ (t) σ· t ρ with: x = + √ 2 σ· t ∗ (7.11 RESULTS AND CONCLUSIONS A company’s credit shortfall risk can be calculated using equations (7.39) The risk-adjusted loan interest rate per annum is then calculated from that. Here the risk must be checked. for an assessment of it to be possible. A further qualiﬁcation emerges on the point of the extent to which information exists at all. as follows: i pa = ∗ i spa + ρpa ∗ 1 − ρpa (7. if important events deﬁned in advance (see above) occur. such as ﬁxed-date mortgage loans over several years. however. it must thus be possible. or the company applying for the loan must be prepared to pay interest on a worst-case scenario basis.37)–(7.40) A loan that has already been granted may be continually assessed for risk-adjustment right through to its maturity.53) The risk-adjusted loan interest rate can also take privileged salary claims into account in the event of bankruptcy: ﬁrst the corrected expectation value of breakdown distributions is . the value should be calculated on an annual basis: ∗ ρpa = 1 − t 1 − ρ ∗ (t) (7.

An Excel worksheet is presented in Appendix 2 with which the equations speciﬁed above may be applied.46). Debts are summarised together as follows: Creditors 1-year loan(s) 3-year loan(s) Total debts 50 500 1000 1550 Debt rate: 1550/2500 = 62% The natural logarithms must ﬁrst be calculated according to equation (7.65) This value must then be converted again according to equation (7.12 EXAMPLES The method outlined here is explained with the aid of two examples.3.1 Example of a Company with Continuous Business Development The key ﬁgures needed for the model are given in Table 7. 0) The corrected credit shortfall risk may then be calculated: ∗ ρc = ρ · 1 − (7.39). The loan interest rate also ensues from equation (7. The liquidation value is lower in all years than the discounted free cash ﬂows. 7. which is why this applies as the company’s value.40).42) for the calculation of volatility: Table 7.3 Key ﬁgures — Example 1 Year Turnover Operating costs Capital investments Free cash ﬂow Discount rate Discounted free cash ﬂows Liquidation value Value of company −3 1000 700 100 200 10% 2000 1000 2000 −2 1025 750 110 165 10% 1650 1000 1650 −1 1100 800 100 200 10% 2000 1000 2000 0 1150 850 90 210 10% 2100 1000 2100 +1 (Budget) 1250 900 100 250 10% 2500 1000 2500 .12.Shortfall Risk on Uncovered Loans: Option-Theory Approach 85 calculated: Bc = Max · (b · L · (1 + i(t)) − S.55) Bc · (1 − ρ) L · (1 + i s (t)) − Bc · ρ (7. 7. according to equation (7.

4.0518% ∗ ρpa = 0.1329% b 93.5 Final results of Example 1 Rating 1 year 3 years AA BB ∗ ρpa according to rating is (assumption) 4.45) a volatility of σ = 19.6.0558 Using equation (7.0488 ln(2500/2100) = 0. The iterative calculation delivers the results shown in Table 7.25% results from this.1924 ln(2000/1650) = 0.0086 0.8554% 11.25% 33. using an Excel worksheet.7570% Table 7.1.0518% 1.1744 mean µ = 0.0518% 1.0763% 5.5724% ρ 0.49 and 7. .50) and shown in Table 7. using equations (7. there is a considerable difference in interest between the one-year and the three-year loan.0050 0.5724% Table 7. The result for ρ ∗ must now be converted to one year.1924 ln(2100/2000) = 0. 7.5269% The one-year loan receives an AA rating and the three-year loan receives a BB rating according to Table 2.3844 −1.48. according to the rules for transforming time periods (equation (7. This leads to the ﬁnal results for the loan rates of interest consistent with risk. It is striking that despite the relatively small volatility of not quite 20% and a debt rate of almost two-thirds.86 Risk-adjusted Lending Conditions Table 7.49) 4.2971% i rounded up 41% 8 5 5 16 % 0.0602 ρ∗ 0.5.0733% 0.0% 4.88% L(1 + i(t)) 520 1159 B 489 996 ln(1650/2000) = −0.34% −2.4 Shortfall risks in Example 1 √ σ t x N (x) 1 year 3 years 19. according to Table 7.5% i according to equation (4.6 Bankruptcy situation in Example 1 ρ∗ 1 year 3 years 0. The situation that may now be expected in any possible case of bankruptcy may be calculated.2195 0.94% 85.1113 √ N (x − σ t) 0.39)): 1 year: 3 years: ∗ ρpa = 0.

65) together with the values from Table 7.9. (7. ratings and loan interest rates shown in Table 7. In year –1 the liquidation value is higher than the discounted free cash ﬂows. The salary and wage claims to be expected in the event of bankruptcy are 62. to the corrected credit shortfall risks.9.49) 4. (Here the interest must be calculated precisely using the values of ρ ∗ given in Table 7. and because the values in the case concerned are correspondingly favourable. what effect the salary claims have on the credit risk. Taking the salary claims in the event of bankruptcy into account in the case of the one-year loan has led to deterioration in the credit-worthiness by one rating level.07 ∗ ρc (%) ∗ ρcpa (%) Rating A BB is (assumption) 4. as in Section 7.7 Proportion of privileged salary claims Liabilities Creditors Loan 1 Year Loan 3 Years Total Amount 50 500 1000 1550 Share of salary claims 2 20 40 62 87 Table 7. Debts are summarised together as follows: Creditors 1-year loan(s) 3-year loan(s) Total debts 50 500 500 1050 .849 0.55).4.7. which is why this value is put in as the company’s value.9959 0. the bank may thus assume that the loans.5% i according to (4. despite an increase in risk.12 82. in the event of bankruptcy according to Section 7. 7.1780% 5. including accumulated interest over the whole terms of them. This comes about because the ‘higher’ rating levels are ‘broader’ than the ‘lower’ rating levels. still have a value of about 94% and after three years a value of about 86%. as shown in Table 7.) It is now our intention to investigate.12.Shortfall Risk on Uncovered Loans: Option-Theory Approach Table 7. This leads.8.2 Example of a Company with a Poor Financial Year The key ﬁgures needed are again given in Table 7.64) and (7. They spread out over the liabilities. The rating level has not deteriorated in the case of the three-year loan.2971% i rounded up 3 4 16 % 5 5 16 % 0.8 Final results after privileged salary claims Bc 1 year 3 years 469 956 bc (%) 90.849 1.7.6698 If the entrepreneur gives up his company after one year.0% 4. using equations (7.9.

2486 ρ∗ 10.8329 ln(1000/1000) = 0.45). The loan rates of interest consistent with risk are substantially higher than in the previous example.47% 118.0000 ln(2000/1000) = 0.1. despite the clearly lower debt.6939 √ N · (x − σ t) 0.10 Shortfall risks in Example 2 √ σ t x N (x) 1 year 3 years 68.47% according to equation (7.39) gives the following results: 1 year: 3 years: ∗ ρpa = 10. .54% Using Table 2.10.6931 mean µ = 0.87% ∗ ρpa = 16. The poor ﬁnancial year − 1 has increased the volatility of the company’s value markedly.8652% Debt rate: 1050/2500 = 52. The transformation of time periods according to equation (7.8658% 41.88 Risk-adjusted Lending Conditions Table 7.3333 0.1323 0. using an Excel worksheet. this leads to the ratings and ﬁnal results for the loan rates of interest consistent with risk shown in Table 7.5% The volatility calculation is made analogously to the previous example: ln(2300/2000) = 0.1398 ln(1000/2300) = −0.11. The iterative calculation delivers the results shown in Table 7.59% −0.9 Starting position Example 2 Year Turnover Operating costs Capital investments Free cash ﬂow Discount rate Discounted free cash ﬂows Liquidation value Value of company −3 1000 700 100 200 10% 2000 1000 2000 −2 1025 750 120 230 10% 2300 1000 2300 −1 1100 800 50 50 10% 500 1000 1000 0 1150 850 100 100 10% 1000 1000 1100 +1 (Budget) 1200 900 100 200 10% 2000 1000 2000 Table 7.5070 0.4307 0.0000 That results in a volatility of 68.

33% 69. in cases of companies having had poor ﬁnancial years in the recent past and whose values are therefore highly volatile.12.0% 4.49) 18. Comparison of the results of the two examples in subsections 7. loan interest rates have been allowed to apply that are much too low.Shortfall Risk on Uncovered Loans: Option-Theory Approach Table 7. .67% L(1 + i(t)) 583 981 B 393 389 The situation shown in Table 7.12.12 is to be expected in any possible event of bankruptcy.87% 41.9871% Table 7. Taking the salary claims in the event of bankruptcy into account by analogy with the previous example is left to the reader.12 Bankruptcy situation in Example 2 ρ∗ 1 year 3 years 10.2 permits the supposition that in today’s banking practice.2993% i rounded up 18 7 % 8 5 39 16 % 12.40% 39.8281% 39.39% b 67.4786% 24.1 and 7.5% i according to equation (4.87% ρ 33.11 Final results of Example 2 Rating 1 year 3 years C DDD ∗ ρpa according to rating 89 is (assumption) 4.

.

The smaller the loan in relation to the value of the collateral.4.3. will be taken into account.5. and the other not. and electronically. How to proceed in the case of a combination of one covered loan and one uncovered loan to the same borrower will be investigated in Section 8. than the market value of the company or the equivalent of that in respect of the private individual. 8. The results of Sections 8. This consideration leads to the same result for the company as in Chapter 7. so long as it is geared to the collateral alone. relax this assumption. no loan may ever be greater. This way of looking at it rests on the assumption that banks usually only make loans when borrowers are. with the consequences which ensue from that fact. with terms being substituted as shown in Table 8.6 by means of an example. at the time of its being issued. This fact is probed in Section 8. is in default if the value of the collateral no longer matches the mortgaging bank’s minimum claim.1. to all intents and purposes. The question of the optimum combination of covered and uncovered loans. and vice versa. Experience shows that this can take from just a few days in the case of securities offered as collateral. The same applies. Collateral here only serves the bank as guarantee against the case that is out of the ordinary. As was illustrated in the comments on Figure 7. in that in the case of the collateral falling short the loan will not necessarily be in default as long as the borrower is still in a position to service it. The results of this chapter will be illustrated in Section 8.8 Loans Covered against Shortfall Risk In this chapter we will examine the shortfall risk of covered loans. In Sections 8.1 and 8.1. on the basis of stock exchange data. will not be investigated until Chapter 9.2 and 8. where the bank is relying exclusively on the cover. More information is thus available for the calculation of . In the case of covered loans the time from the completion of the loan to the realisation of the collateral must also be taken into account in the bank’s reaction time.2 the correlation between the probabilities of the collateral falling short and of the borrower defaulting will ﬁrst be examined. the smaller is the risk on the loan. in passing. but on which the bank relies on the collateral alone: the loan may not be higher than the value of the collateral. In the case of securities it is possible to reassess the collateral daily. analogously. in a position to service them without any difﬁculty. for loans that are covered. where the bank in principle relies exclusively on the collateral. Here only the fact that one loan is covered. In Section 8.3 we will.2 will be summarised in Section 8.1 SHORTFALL RISK OF A COVERED LOAN ON THE BASIS OF THE OPTION-THEORY APPROACH A covered loan. and thus also the question of a loan that is partially covered. to several years in the case of mortgages.

27) [CORU85.38).92 Risk-adjusted Lending Conditions Table 8. They are therefore detailed once more at this juncture: ∗ pC √ NC (xC − σC · t) − dC · NC (xC ) = √ NC (xC − σC · t) − dC (7.37) .37)–(7. S. The determination of the current value of a property is common practice for any bank. but also makes it possible for an individual calculation of the probable development in the value of any property to be drawn up. This method not only allows for the published index to be drawn up. the current value of the property concerned must be known. Credit is due to the Cantonal Bank of Z¨ rich’s pioneering work in this ﬁeld. L E σ d t Chapter 7 Market value of the company Debts Equity Volatility of the value of the company Debt rate (Remaining)term of loan and/or reaction time in the case of current account lending Chapter 8 Value of collateral Mortgage Unmortgaged portion of collateral Volatility of the value of the collateral Mortgage rate (Remaining)term of loan and/or reaction time in the case of current account lending and variable mortgages volatility in the case of securities used as collateral. but they are known and largely under control. together with the volatility of this value.1) ln k=1 Xk Xk − 1 X = portfolio values A = number of trading days per year In order to be able to assess mortgage loans. It cannot actually be said that this is free of problems. It is important to distinguish clearly between these values. The loan risk of the covered loan is calculated using equations (7. It is more difﬁcult with the determination of volatility. For this not only the current value of any property but also the probable development in its value has to be known. 256]: σ = with µ = A · n−1 1 · n n n ln k=1 Xk Xk − 1 2 −µ (8.48) and the breakdown distribution probability value using equation (7. See [ZKB96] for details. the shortfall risk of the collateral using equation (7. which leads to a simpliﬁcation vis-` -vis a equation (7.1 Uncovered versus covered loans Variable V D. Index valuations may be downloaded from the Internet (see Appendix 3). whereby it developed a property u price index based on the ‘hedonistic’ method.49).

that neither the borrower defaults nor the collateral falls short. It is still of course just as much as before the bank’s contractual partner. The borrower may indeed be in as good a position as previously to meet its obligations to the bank on the strength of its solvency. that only the collateral falls short. ˆ r Probability b.48) bC = 1 − Index C = collateral ρC = ∗ ρC NC (xC ) 8.38) (7.1 . This section will therefore deal with determining the combined shortfall risk of the borrower and of the collateral. 8. but that the collateral does not fall short. that both the borrower defaults and the collateral falls short. ˆ r Probability d.1. The ‘no default’ occurrence has a value of ‘0’. that does not necessarily mean that the loan is effectively lost to the bank. ˆ r Probability c.Loans Covered against Shortfall Risk 93 with x = √ + σ· t ρC = NC (xC ) ∗ ρC ln d 1−ρ ∗ √ σ· t 2 (7. but that the borrower does not default. taking the correlation between the two into account.47) (7. ˆ This can be illustrated graphically as shown in Figure 8. that the borrower defaults.1 Derivation of the Correlation The probabilities of four possible occurrences within a period of time have to be considered in the case of a covered loan: r Probability a. collateral The ‘default’ occurrence has a value of ‘1’.2.2 CORRELATION BETWEEN THE SHORTFALL RISK OF THE BORROWER AND THE SHORTFALL RISK OF THE COLLATERAL If a covered loan defaults only on account of the collateral. 1 0 c b 0 a d borrower 1 Figure 8.

whereby this involves a theoretically conceivable perfect hedge that is. in which there is no correlation at all. S.1.6) . 304] for the correlation of this frequency distribution: ˆ ˆ ˆ ˆ ˆ a − (a + d) · (a + c) ˆ (8.3) r= ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ [(a + d) − (a + d)2 ] · [(a + c) − (a + c)2 ] ˆ r = correlation coefﬁcient After insertion of the shortfall risks according to equation (8. S.94 Risk-adjusted Lending Conditions The following correlations are discernible on the basis of Figure 8. as one would expect according to the multiplication rules that apply to the probability of independent occurrences [BOHL92. provided the individual shortfall ˆ risks and the correlation coefﬁcient are known. 8. i. unrealistic in practice. ρB ∩ C = 0 ˆ ρB = d ˆ ρC = c ˆ ˆ ˆ 1 = b+c+d (8. equation (8.2).2.e. ˆ ρB ∩ C = a ˆ ˆ ρB = a + d ˆ ˆ ρC = a + c ˆ ˆ ˆ ˆ 1 = a+b+c+d (8. 324].2) ρ B ∩ C = combined risk Index B = borrower One obtains the following [KREY91.2 Value Area of the Efﬁciency of the Correlation ˆ It is intended next to investigate what values may theoretically be assumed for r and ρ B ∩ C . one obtains: ˆ r= Solution using ρ B ∩ C results in: ˆ ρ B ∩ C = ρ B · ρC + r · 2 2 ρ B − ρ B · ρC − ρC ρ B ∩ C − ρ B · ρC 2 2 ρ B − ρ B · ρC − ρC (8. however.5) reduces itself to ρ B ∩ C = ρ B · ρC . Minimum First of all the minimum should be calculated.2. The frequency distribution may then be portrayed as follows and shown graphically in Figure 8.5) It is then possible to calculate the combined shortfall risk.4) (8. In the special case r = 0.

Loans Covered against Shortfall Risk 95 collateral The ‘default’ occurrence has a value of ‘1’. 1 0 c b 0 0 d borrower 1 Figure 8. on the basis of equation (8.235].6). Maximum ˆ Maximum values for r and ρ B ∩ C are reached by the consideration that such is the case if collateral falling short leads to default on the loan.12) ˆ r may thus only assume the value −1 if the sum of both individual risks is equal to 1.9) 2 2 ρ B − ρ B · ρC − ρC Conversion results in: 2 2 2 2 ρ B − ρ B · ρC − ρC = ρ B · ρC (8. −ρ B · ρC −1 = (8.7) ˆ ˆ (d − d 2 ) · (ˆ − c2 ) c ˆ After insertion of the shortfall risks according to equation (8. We now intend to investigate under what assumptions this value may be reached. S. the lowest value that a correlation coefﬁcient may assume is −1 [BOHL92.10) Multiplying out gives: 2 2 2 2 2 2 ρ B · ρC − ρ B · ρC + ρC · ρ B + ρ B · ρC = ρ B · ρC (8.11) Which leads to: 1 = ρ B + ρC (8. that b = 0 in this case.6).2 One obtains the following [KREY91. S.e. 304] for the correlation of this frequency distribution: ˆ ˆ 0−d ·c ˆ r= (8.8) According to statistical theory. ρ B ∩ C = ρC . This ˆ means. i. The ‘no default’ occurrence has a value of ‘0’. one obtains: ˆ rmin = −ρ B · ρC 2 ρ B − ρ B · ρC − ρ 2 C (8. The frequency distribution .

This makes sense. ˆ ρB ∩ C = a ˆ ˆ ρB = a + d ˆ ρC = a ˆ ˆ ˆ 1 = a+b+d (8. this means that ρ B > ρC .12).3. as the bank would of course not otherwise fall back on collateral at the time of granting the loan. S. S.18) . The ‘no default’ occurrence has a value of ‘0’. 1 0 0 a d borrower 1 b 0 Figure 8.17) Multiplying out gives: 2 2 2 2 2 2 2 2 ρ B · ρC − ρ B · ρC − ρC · ρ B + ρ B · ρC = ρC − 2 · ρ B · ρC + ρ B · ρC (8. One obtains the following [KREY91. 304] for the correlation of this frequency distribution: ˆ rmax = ˆ ˆ ˆ ˆ a − (a + d) · a ˆ ˆ ˆ ˆ ˆ ˆ [(a + d) − (a + d)2 ] · [a − a 2 ] ρC − ρ B · ρC 2 2 ρ B − ρ B · ρC − ρC (8.13) ˆ ˆ As d ³ 0 applies as probability value for d. one obtains: ˆ rmax = (8. The shortfall risk of the borrower is thus greater or at least equal to the shortfall risk of the collateral and to the combined shortfall risk respectively.16) (8. 1= Conversion results in: 2 2 ρ B − ρ B · ρC − ρC = (ρC − ρ B · ρC )2 ρC − ρ B · ρC 2 2 ρ B − ρ B · ρC − ρC (8.96 Risk-adjusted Lending Conditions collateral The ‘default’ occurrence has a value of ‘1’.14) After insertion of the shortfall risks according to equation (8. the highest value that a correlation coefﬁcient may assume is +1 [BOHL92.15) According to statistical theory.3 can then be portrayed as follows and illustrated in Figure 8.235]. We intend therefore to investigate under what assumptions this value may be reached.

(ρ B ∩ C )∗ = (ρ B ∩ C ) · (1 − bC ) By insertion of equations (8. This was often the case in the ﬁnancing of owner-occupied houses in the middle .41) one obtains: ˆ (ρ B ∩ C )∗ = ρ B · ρC + r · 2 2 ρ B − ρ B · ρC − ρC ∗ ρC ρC (8.23).19) Taking out of brackets and abbreviation gives: ρ B = ρC (8.21) · (8.24) is therefore permissible.e. In the case of a correlation close to zero it may thus well occur that the value of the collateral falls below the nominal amount of the loan. in the case of loans secured against collateral in the form of securities).23) In the special case of the maximum correlation.24) ρ B must be determined according to the rules from Section 7. 8. This means. This does. which is indeed a prerequisite for default occurring on a covered loan. Equation (8. equation (8.13). but not the realisation of the borrower’s other asset values.22) and (8. i.4 for applying equations (8. in which there is no correlation at all.5) details the risk that both the borrower defaults and the collateral falls short.23) and (8.22) reduces itself. ˆ on the basis of equation (8. in order to obtain the shortfall risk of the covered loan. ˆ In practice the challenge consists above all in determining the correlation coefﬁcient r for the various loan transactions empirically.3 SHORTFALL RISK OF THE COVERED LOAN This section is concerned with bringing together the individual elements that have been calculated to date. using statistical methods. Borrower default and collateral fall short therefore always occur either simultaneously or not at all. however. make sense and is in line with the principle of conservatism. This probability must be multiplied by the risk of loss (1 − bC ). Here one may put forward the supposition that the correlation coefﬁcient for many loan transactions lies either close to zero (for instance. equation (8. In proceeding thus it is implicitly assumed that only the realisation of the collateral makes breakdown distributions possible. after insertion of equation (8. and the application of equations (8.20) ˆ r may thus only assume the value +1 if both individual risks are equally large.Loans Covered against Shortfall Risk 97 Which leads to: 2 ρ B − ρ B − ρC + ρ B · ρC = 0 (8.22) ˆ In the special case r = 0. that d = 0.5) and (7.15) to: ∗ (ρ B ∩ C )∗ = ρC ˆ ˆ if r = rmax (8. but that the borrower continues to meet his obligations.20) reduces itself to: ∗ (ρ B ∩ C )∗ = ρ B · ρC ˆ if r = 0 (8. in the case of ﬁnancing owner-occupied houses) or close to the maximum (for example.

This action is consistent with the principle of conservatism. The suppositions put forward here are indeed plausible. The claim that is covered must then be deducted from this total breakdown distribution probability value. the simplest case is assumed. in which a company’s debt consists merely of one covered and one uncovered loan. that the preferential demand will be met in full.48). First of all. (7. It is important in such a situation that bank does not lose its nerve and call in the loan unnecessarily. only a few borrowers are in a position to continue to service the loan out of other income. the credit risk in relation to the loan’s nominal value may be calculated according to equation (7. The correlation coefﬁcient in such cases usually lies close to the maximum.49). In assessing the uncovered loan the procedure has to be analogous to that described in Section 7. Portraying bankruptcy proceedings in the form of a model forms another step in the assessment of the loans. experience is quite different. equation (8. ones that are affordable by the borrower).4 COVERED AND UNCOVERED LOANS TO THE SAME BORROWER Both covered and uncovered loans are regularly granted to companies simultaneously. On the basis of the breakdown distribution probability value of the uncovered loan calculated in this way. First of all the breakdown distribution probability value has to be calculated in relation to the whole amount of loan: B = 1− ρ ∗ (L) ρ(L) · L · (1 + i tot ) cf. until the value of the collateral has returned to stand at a ‘reasonable’ level in relation to the amount of the loan. in the worst-case scenario as far as it is concerned. but the borrower continued to have the same income as he had previously. This procedure will be illustrated by means of an example in the next section but one. We have to proceed analogously in the case of debts of complicated structures.9.22) may be applied directly. If the value of the portfolio of securities that has been mortgaged falls below the loan’s nominal value. The principle is always the same: the breakdown distribution probability values have to be determined ﬁrst in total. . It is simpler here to assess the covered loan. followed by scheduling allocation of the breakdown distributions according to the Swiss laws on the recovery of debt and on bankruptcy. Because it has preferential status when it comes to bankruptcy. In the case of loans secured against collateral in the form of securities. The bank that grants the uncovered loan must of course assume. The question thus arises when assessing the individual loans of how to analyse the facts about the company. 8.e. It is better to proﬁt from the low correlation by requesting moderate additional repayments (i.98 Risk-adjusted Lending Conditions of the 1990s: the value of the owner-occupied house had fallen sharply in the course of the general crisis in the property market. but have yet to be corroborated empirically. Let us explain below what is meant by this.50) in which L corresponds here to the sum of the covered and uncovered loans. The determination of the company’s shortfall risk itself is made using equation (7.

22) ˆ As we have demonstrated.23) (8. If. simpler expressions: ∗ ˆ (ρ B ∩ C )∗ = ρ B · ρC if r = 0 ∗ ∗ ˆ ˆ (ρ B ∩ C ) = ρC if r = rmax (8. 8. The necessary substitutions have to be made for this. Even if a loan’s collateral does fall short.24) ˆ Here rmax is not necessarily equal to +1! That is only the case if ρ B = ρC . We shall work through an example in the next section. The shortfall risk of the mortgage combined with the shortfall risk of the company is ﬁrst calculated on the basis of this starting position. very high in the case of loans with securities as collateral.6. Just the calculation of the volatility of securities accounts is calculated according to a slightly modiﬁed equation (8. the shortfall risk of the collateral is greater than the shortfall risk of the borrower. this does not necessarily mean that there will be default on the loan. This will be elaborated in Chapter 9. The procedure is described in the preceding section. in fact in Section 9. privileged wages and salary claims 3.6 the shortfall risk of . mortgage 2. A mortgage shortfall risk of 0. This consideration leads to the following. it is plausible to assume that the correlation coefﬁcient r is very small in the case of mortgages and.1% has been calculated according to equations (7. therefore.37)–(7. The single difference consists in it involving a three-year ﬁxed mortgage on a loan of three years.1. in real life.12. The borrower may nonetheless be in a position to service it.38) in Section 8. Combinations of covered and uncovered loans to the same borrower may be calculated.5 RESULTS AND CONCLUSIONS Loans where the bank relies solely on the collateral may be calculated using the same calculation rules as uncovered loans to companies.1). in contrast. of a covered and an uncovered loan.6 EXAMPLE The same data is applied for this example as in subsection 7. The allocation of breakdown distributions under the Swiss law on debt recovery and bankruptcy is: 1. The shortfall risk is calculated in this case as follows: ˆ (ρ B ∩ C )∗ = ρ B · ρC + r · 2 2 ρ B − ρ B · ρC − ρC · ∗ ρC ρC (8.1. According to Table 7.Loans Covered against Shortfall Risk 99 8. uncovered loans The correlation between the company’s shortfall risk and the shortfall risk of the collateral may be set at zero.1. as listed in Section 8. then a partially covered loan should be calculated: the combination.

) One then obtains a corrected loan risk of 0. the bank will receive a higher proportion of the breakdown distribution probability values. This produces a corrected breakdown distribution rate of 62. this results — using equation (4. the Table 8. Comparison of the results of subsection 7.1 with a rating risk of 0. but the error arising is small and leads to a slightly increased risk.94% according to Table 7.13%. The probable breakdown distribution rate expected in one year comes to 93. or rounded up to 4 16 %.3824%. This is on balance better both for the borrower and the bank: the borrower is paying loan interest reduced by 7 and the bank has lower loan risk on account of the collateral.1113 · 0. the creditors are also considered as claims on which interest may be claimed. ensues using equation (4. The preferential mortgage demand.2 Final results Term 3-year loan without collateral 3-year loan with collateral 1 year 3 years AA BB 41% 8 5 5 16 % BBB AAA 7 4 16 % 9 4 16 % . as one would expect.5% (Table 7.12.49). including the two 9 years of interest to be expected of 4 16 % of 1093.74% in the case of uncovered claims of 550. This is indeed not quite correct. following the deposit of collateral. This is acceptable in the light of the principle of conservatism. falling to 67 following doing so. This is in line with a BBB rating. This results in probable breakdown distributions of 360 for the non-preferential claims. The company was paying annual interest of a total of 74 prior to depositing collateral. The value of 360 for BC may be inserted into equations (7. together with the privileged wages and salary claims of 62. If the risk-free rate of interest i s is 4% (Table 7. The lower risk for the bank ensues above all from the fact that the breakdown distribution probability values for the creditors are smaller than they were previously. This breakdown distribution is now allocated proportionately to the uncovered creditors of 50 and to the uncovered loan of 500 plus accumulated interest. this yields a breakdown distribution probability value of 1515.011113% This corresponds to an AAA rating as per Table 2. according to equation (8.2) gives a picture of the loan interest rate and the rating. The assessment of the uncovered loan takes place in a second step.6.5).001 = 0. which in turn takes the form of a lower total interest expectation. r continues to be equal to zero. have to be deducted from this value.5).0244%. however.65) for further calculations.1.5).1 and this section (Table 8. If the risk-free rate of interest i s = 4.49) — in a 9 mortgage rate of interest of 4.61–7. the rating of the one-year loan has deteriorated markedly. or rounded up to 4 16 %.2648%.5255%.23): (ρ B ∩ C )∗ = 0. Conversely.08% (Table 7. a 7 loan interest rate of 4. Because of the collateral. Related to the total debt of 1550 plus accumulated interest of 4. (For the sake of simplicity.3663% according to Table 2.100 Risk-adjusted Lending Conditions ˆ the company over three years is ρ B = 11. The rating of the three-year loan has improved markedly. The rating risk is 0.

It may be assumed.Loans Covered against Shortfall Risk 101 total of which will remain the same. the situation will deteriorate considerably for the other. at the expense of the creditors. as one might expect. in so far as uncovered loans are granted. a restrictive clause on mortgaging in the loan agreement. unless it is immediately informed of the new situation and can adjust its own conditions directly to the new circumstances. additionally. that two different banks grant the two loans. The example shows clearly that in the case of borrowers that seek loans from several banks. After the depositing of collateral in favour of one bank. . The depositing of collateral is. it is absolutely essential to have.

.

1.1 MARGINAL INTEREST RATE As can be seen in the ﬁgures in Section 7. into: i m = lim i(L + L) · L + i(L + L) · L L − i(L) · L (9. and illustrated in Section 9. and in Section 9. the maximum degree of debt. In determining such ﬁnance for a company.8. however.9 with the aid of an example. A central theme in the granting of loans is the acceptability of the debt servicing and. 9. as the following shows: i m = lim i(L + L) · (L + L) − i(L) · L L (9. The results and conclusions will be summarised in Section 9. which is why this subject is dealt with in Section 9.2) L→0 . The additional interest costs do. the marginal interest rate of a loan is of central importance.5. associated with that.9 Calculation of the Combination of Loans with the Lowest Interest Costs In Chapter 8 we have shown how one must proceed if a company is simultaneously seeking covered and uncovered loans. Building up to the cases of two (Section 9.1) L→0 The marginal rate of interest thus corresponds to that rate of interest that is received. as a result of multiplying out.1) turns. This can be portrayed graphically as shown in Figure 9. any increase in the debt rate or mortgaging has the effect of an increase in the loan interest rate for the whole loan.6.2) and three (Section 9. only arise on account of the increase itself and thus correspond in principle to the interest costs of the increase.3) loans. the rules are derived for whatever number of loans one likes (Section 9.1 at the beginning of this chapter. at what ratio of debt to equity the highest returns on equity may be achieved. if one allocates the entirety of additional interest costs arising from an inﬁnitesimally small increase in loan to that increase. This chapter will be concerned with laying down that combination of different loans which gives rise to the lowest possible loan costs for any company. i m is deﬁned as the marginal interest rate. This theme will be covered in Section 9.5 how partially covered loans may be calculated.7 on the strength of our ﬁndings regarding the most favourable ﬁnancing in terms of interest costs. Equation (9.4) that lead to the ﬁnancing that is the most favourable in terms of interest costs. We will demonstrate in Section 9.

however.49) is inserted into equation (9. the gradient of the straight lines through points AB is less than the gradient of the interest curve at point A. a simple possibility of determining the marginal interest rate approximately. In the case of the second limit.4) is preserved.4) The marginal rate of interest of a loan amounting to L is thus the sum of the rate of interest of this loan together with the partial differential of the rate of interest. according to that.4). runs as follows: i m = i(L) + L · ∂ i(L) ∂L (9. in that it is shortened by L. The solution. As explained in Section 7. In principle this involves determining the gradient of the interest curve at point A. the credit shortfall risk in equation (9. in that in addition to the rate of interest i(L).1 L + ∆L Transpositions yield: i m = lim i(L + L→0 L) · L L + L · lim i(L + L→0 L) − i(L) L (9.2). To anticipate the reader’s question immediately — it is not possible to do so algebraically! If equation (4. The problem emerges here of determining this partial differential. Thus there is no other solution than to solve this differential iteratively too.104 Risk-adjusted Lending Conditions i i (L + ∆L) i (L ) is L L Figure 9. There is. The gradient of the . As can be seen in the ﬁgure. the credit shortfall risk has to be solved iteratively. it is a question of a differential quotient.3. On the other hand the gradient of the straight lines through points AC is greater than the gradient of the interest curve at point A.3) The solution of the ﬁrst limit is insigniﬁcant. the two rates of interest i(L + L) and i(L − L) are calculated (see Figure 9. derived from the amount of the loan and multiplied by the amount of the loan.

however. substantially steeper than that of i. such that L = L 1 + L 2 . The maximum absolute error that is made here corresponds to the difference in the slope of the straight lines AC and BC. which means that the following applies: i 1 · L 1 + i 2 · L 2 = min with L = L 1 + L 2 (9. 9.6) . The approximate value for the marginal rate of interest thus works out as: i m = i(L) + L · i(L + L) − i(L − 2· L L) (9.2 TWO LOANS Every borrower’s aim is to have the lowest possible interest costs. The gradient of the straight lines though points BC may thus be selected as an approximate value for the gradient of the interest curve at point A. Let us assume that a borrower needs outside capital to the extent of L. The rise of i m is.3 shows the correlation between interest rate i and marginal interest rate i m . which can be calculated.Combination of Loans 105 i i (L + ∆L) i (L ) i (L − ∆L) is A B C L L − ∆L Figure 9. It is striking that i and i m may be approximately identical for a large ﬁeld of values of d. one covered and one uncovered). The value of L must thus be chosen so that this difference is sufﬁciently small for the corresponding application of the result.2 L L + ∆L interest curve at point A thus lies between the gradients of the two straight lines described. The same applies for the gradient of the straight lines through points BC. This sum may be divided into two individual loans L 1 and L 2 (for example. from a speciﬁed value of d onward.5) Figure 9.

7) Thus the lowest interest costs arise when the overall amount of loan is so divided that the two marginal interest rates are identical.6) at L 1 results in zero: ∂ [i 1 · L 1 + (L − L 1 ) · i 2 (L − L 1 )] = 0 ∂ L1 According to the rules of differential calculus.3 By expressing L 2 as L − L 1 .10) (9. the variable L 1 now has to be determined in such a way that minimum interest costs arise. .8) (9. as otherwise part of any loan could be replaced by making an increase in the other loan on more favourable conditions.2.4) this is none other than: i m1 = i m2 (9. That is plausible too. one obtains: i 1 (L 1 ) + L 1 · i 1 (L 1 ) − i 2 (L − L 1 ) − i 2 (L − L 1 ) = 0 Reverse substitution of L 2 = L − L 1 gives: i 1 (L 1 ) + L 1 · i 1 (L 1 ) − i 2 (L 2 ) − L 2 · i 2 (L 2 ) = 0 But from equation (9.9) (9.106 Risk-adjusted Lending Conditions i im 25% 20% i 15% 10% 5% is 0% 0% d is = 3%. L 1 has to be chosen in such a way that the ﬁrst derivation from equation (9. 50% = 0. t = 1 100% Figure 9.

Combination of Loans 107 9.3 THREE LOANS This is done in the same way as in the preceding section.15) (9.16) (9. 9. We will thus investigate the situation in which a borrower seeks n loans and wishes thereby to achieve interest costs that are as low as possible.4 THE GENERAL CASE OF SEVERAL LOANS This involves generalising the ﬁndings of the two preceding sections. L n−1 .8) is written here as follows: i 1 · L 1 + i 2 · L 2 + i 3 · L 3 = min with L = L 1 + L 2 + L 3 (9.12) According to the rules of differential calculus.14) (9. L 1 and L 2 . First the partial differential following L 1 is formed: ∂ [i 1 · L 1 + i 2 · L 2 + (L − L 1 − L 2 ) · i 3 (L − L 1 − L 2 )] = 0 ∂ L1 (9. The following therefore applies: n n−1 L j · i j = min j=1 with L n = L − j=1 Lj (9.17) There are now n − 1 variables L 1 . .11) Now there are two variables. one obtains: i 1 (L 1 ) + L 1 · i 1 (L 1 ) − i 3 (L 3 ) − L 3 · i 3 (L 3 ) = 0 This again gives: i m1 = i m3 The partial differentiation following L 2 ensues. the ﬁnancing that is the most favourable in terms of costs is attained when all marginal interest rates are identical. and these may be written in the following way: ∂ ∂ Lk · L k · i k (L k ) + n−1 j=1 j=k L j · i j L j + L − L k − n−1 j=1 j=k L j · in L − L k − n−1 j=1 j=k L j = 0 (9. N − 1 partial differentials have therefore to be formed.13) In this case. That means there are two partial differentials to be formed following L 1 and L 2 . Equation (9. too. which must both be equal to zero. . analogously: i m2 = i m3 This leads to the ﬁnal result: i m1 = i m2 = i m3 (9. and following reverse substitution of (L − L 1 − L 2 ) = L 3 .18) .

that a company has the most favourable outside ﬁnance in terms of interest costs when all marginal interest rates are identical. on the basis of the proportion that each bears to the total loan. This may make perfectly good sense in certain situations. even if the return on overall assets is volatile in itself and in relation to itself. First of all therefore the division between the two part loans that is the most favourable in terms of interest costs must be determined according to Section 9.e. none other than the combination of a covered and of an uncovered loan. i. dependent on the term of the loan.108 Risk-adjusted Lending Conditions According to the rules of differential calculus. one obtains: i k (L k ) + L k · i k (L k ) − i n (L n ) − i n (L n ) = 0 for all k with 1 £ k £ n − 1 But this is again none other than: i mk = i mn for all k with 1 ≤ k ≤ n − 1 (9. Then each part loan must be assessed according to Chapters 7 and 8. as the conditions of the other loans are bound to have been ﬁrmly agreed. 9. but we will not go into the reasons here. then it must be taken into account that the optimisation may no longer necessarily be attained at the time of the next extension of a loan. How to do that will be demonstrated below.20) (9. at the end of the day. where it will be assumed that the long-term average return on assets is ﬁxed. The rate of interest for the loan as a whole is then worked out at the end as the weighted average of the rates of interest for the two part loans.2 for the calculation of this loan. a more stable average may result. One means to that end is the optimisation of the ratio between debt and equity. the value of the collateral is lower than the amount of the loan.5 PARTIALLY COVERED LOANS It is frequently the case in banking practice that loans to companies are partially covered. in the meaning of the theory described here. 9. The optimum solution in such situations has to be found by using model calculations on the basis of various scenarios. It is worth reﬂecting at this point that the marginal interest rates are. A partially covered loan is. and the risk-adjusted rate of interest worked out.19) So it also applies in the general case.6 MAXIMUM RETURN ON EQUITY All entrepreneurs strive to achieve maximum return on equity invested. If loans with different terms are now optimised with each other. The following new variables are introduced: return on assets: return rate: return on equity: v = EBIT/V g = company proﬁt/V e = company proﬁt/E . There is no inconsistency here. As better years and worse years succeed each other.

The general case with n loans can be derived likewise.29) . plus equity capital. one obtains: −(i (d) · d + i(d)) · (1 − d) + v − i(d) · d = 0 Solving by v one obtains: v = i(d) · d + (i (d) · d + i(d)) · (1 − d) This is.21) v − i(d) · d g = (9. however: v = i(d) · d + i m (d) · (1 − d) (9. the debt rate d must selected in such a way that the differential of equation (9. if the debt rate is selected in such a way that the interest costs. one obtains: n n −[i 1 d1 · d1 + i 1 (d1 )] · 1 − d1 − j=2 dj + v − i 1 (d1 ) · d1 − j=2 i j (d j ) · d j = 0 (9.Combination of Loans 109 On the basis of these deﬁnitions we can write: g = v − i(d) · d (9.26) means precisely that any company has a debt/equity ratio that leads to a maximum return on equity.27) dj 1− j=1 Whereby the debt rate was analysed into the n parts of the individual loan d1 to dn .22) from d is equal to zero: ∂ ∂d v − i(d) · d 1−d =0 (9. In individual cases this condition may again only be solved iteratively. The ﬁrst runs as follows: n v − i 1 (d1 ) · d1 − i j (d j ) · d j ∂ j=2 (9. On the basis of the derivation the above statement applies in principle only for one company with one loan. multiplied by the marginal interest rate of the debt structure that has the most favourable interest costs. corresponds to its EBIT.24) Equation (9. Now n partial differentials have to be formed.23) According to the rules of differential calculus.25) (9.26) (9.22) e= 1−d 1−d In order to attain the maximum return on equity e. The return on equity now amounts to: v− e= n j=1 i(d j ) · d j n (9.28) =0 n ∂d1 1 − d1 − dj j=2 According to the rules of differential calculus.

4). It was demonstrated in the preceding section (9.4 Solving by v and abbreviated with d = d1 + n d j one obtains: j=2 n v = i m1 · (1 − d) + j=1 i j (d j ) · d j (9. according to the preceding section (9. 0. v = 10%.30) may thus be replaced by the general marginal interest rate i m .1 0. must be inserted as the marginal rate of interest.4 = 0. and the second summand corresponds to the average rate of interest on debt. The marginal interest rate i m1 in equation (9. t = 1.110 Risk-adjusted Lending Conditions d/ ( e = i m ) 100% i/ 10% d i e=im 50% 5% 0% 0 0.31) dj 1− j=1 . The rate of interest that provides ﬁnance at the most favourable rate in terms of interest costs. The maximum return on equity may thus be calculated as: v− e= n j=1 i j (d j ) · d j n (9. when marginal interest rates are identical for all loans.2 0.4) that the debt is structured at its most favourable in terms of interest costs.3 is = 3%.5 Figure 9.30) The ﬁrst summand is again the equity rate multiplied by the marginal interest rate.2 0.

81. Figure 9..96% 30% 20% e i 10% is 0% 0% 50% EBIT = 10%.4 and 9. t = 1 100% d Figure 9. and not the balance sheet value.5 elucidate this correlation for any company with one loan.8% = 0. also make it clear that the most important objective for any company must be to attain the lowest possible . Figures 9. The graphs.30) this is: i m · (1 − d) + e= n j=1 i j (d j ) · d j − (1 − d) n j=1 i j (d j ) · d j = im (9.Combination of Loans 111 im 30.4 shows that high returns on equity may be achieved when the volatility of the value of the company is low. has been used for the value of companies in preceding chapters.2. It may be recalled at this juncture that the market value. The statements made in this section in relation to equity and return on equity must therefore be related to the market value: equity = market value less debt Maximum return on equity has therefore been derived on equity as deﬁned above. is = 3%.e. even if the debt rate selected is very high and even if the credit shortfall risk and loan interest rate are correspondingly high. however. and not on equity as deﬁned by the company’s books of account. the maximum achievable return on equity corresponds to the marginal interest rate of the debt structure that is the most favourable in terms of interest costs in the case of any appropriate ratio of debt to equity.32) i.5 According to equation (9.

in the case of return on assets being equal. It follows from this that. If a company is ﬁnanced by loans over different terms. with as little volatility as possible.1 Acceptability of Interest Rates In the preceding section we have demonstrated how the ﬁnancing structure of a company may be determined with the maximum achievable return on equity.5 that it is dangerous to ﬁnance a company in such a way that maximum return on equity is achieved. . It obviously makes no sense at all to ﬁnance any company at a high debt rate. if the risk-adjusted debt servicing that results from it leads straight to bankruptcy.4).7 ACCEPTABILITY OF DEBT SERVICING In the preceding chapters we have demonstrated how risk-adjusted loan interest rates may be calculated. it is absolutely essential for any company to keep revenues — and thus its market value — at the highest possible level. than a comparable company. Even small increases in the debt rate lead in this situation to sharply reduced returns on equity. We must be clear here that the return on assets has been calculated on the company’s market value and not on its assets as deﬁned in its books of account. Conclusions The marginal interest rate i m is always larger or equal to the rate of interest i. also comments on Figure 7. The quintessence of this section results in the following: on the assumption that all banks come to apply risk-adjusted loan conditions as described in this study. the acceptability of the interest is automatic. then the capital structure that leads to maximum return on equity should be investigated by model calculations. It is discernible from the course of return on equity e in Figure 9. If this condition is fulﬁlled. any company the market value of which has higher volatility has a lower debt rate leading to a maximum return on equity. High volatility in the company’s market value and high rates of debt mean high interest rates.26) to be fulﬁlled. Here due heed should be given to the notion of safety. and vice versa. and the marginal interest rate and return on equity respectively higher than return on assets. 9. The most important conclusion at that point is the following: the loan interest rates of a company ﬁnanced from the point of view of maximum achievable return on equity are lower than its return on assets. 9. in order to achieve a high return on equity via a high level of debt made possible thereby. For the condition of equation (9. What has until now been left out of consideration is whether the interest rates so calculated are in general acceptable for companies.112 Risk-adjusted Lending Conditions volatility in its market value. and vice versa. the rate of interest i must therefore be lower than the return on assets. in order to achieve the highest possible return on equity with high debt rates (cf.7.8). with the aid of different scenarios (see preceding Section 9.

because the bank has. such that they are acceptable to the company. the proﬁt on equity of 20% of the company’s market value amounts to about 6% of the company’s market value. Just one small but sharp drop in earnings and the ensuing drop in market value. Let us take for granted that the debt rate has risen to 80% as a result of new investment. it makes no sense for a bank to demand repayments — by doing so it would indeed be passing up. In the case of a debt rate of 90%.2 Acceptability of Repayment When it comes to repayments. It is evident from the course of the e-curve that even very small increases in the debt rate lead to a rapid falling away of return on equity. the company achieves a maximum return on equity if the debt rate is about 82%. to below 70% and is then probably back again in the area that is safe. The following is now relevant as far as repayments are concerned. reverting to the example in Figure 9. In view of the danger outlined above. The return on equity then amounts to about 30%. under which it might come close to the point of being ﬁnanced at the level at which return on equity is maximised. In the example selected there. The upper limit here is reached if the company’s owners pass up dividend payouts completely. to raise the risk-adjusted loan interest rate sharply. i. This was waived for the sake of simplicity. (To be correct. in this example taxation should also have been taken into account.e. the optimum debt rate in the example illustrated in Figure 9. It makes sense in such situations for the bank to agree on repayments that in turn lead the company away from the ﬁnancial point at which such dangers arise. run into the dangers outlined above.e. and the increase in debt rate arising therefrom leads immediately to ﬁnancing on the basis of sharply reduced return on equity. The interest rate on the debt amounts at this point to about 5%. good business. The return on equity then is always still between 20% and 30%.7. The consequence of the course of the e-curve is therefore that it is dangerous to ﬁnance a company to the level that matches maximum return on equity. however. It follows from this that the company could actually reduce the debt rate from 80% to 74% within one year. then the repayment is reduced to 4% of the company’s market value. On the other hand this then means also that there is nothing left there at all. other things being equal.Combination of Loans 113 9. If the owners of the company are entitled to 10% dividend payout. and may even lead to a case for winding up. the question that arises ﬁrst and foremost is why repayments should be demanded from the bank at all. It may be instructed to make an increase in its debt rate. Then all of the free cash ﬂow remaining after interest and taxation may be applied to repayments. at least partially. The answer to this question is illustrated in Figure 9. proﬁtable. i.5. In doing so it would. The following may now be relevant.5 lies somewhere between 60% and 80%. as a countermove.5. Let us take for granted that any good. of about 10%. the risk-adjusted loan interest rate is doubled to almost 10% and the return on equity sinks by about two-thirds to the same ﬁgure. Then debt goes down within three years. It may be argued that so long as a company is paying risk-adjusted interest. according to the i-curve. and a return on equity of about 31% results from that.) . company wishes to expand and has to invest accordingly. So the question that comes up now is at what level repayments should be set by the bank.

The consequences resulting for the bank are: r increasing the loan interest rate r demanding repayments.5 Consequences for Loan Supervision Derived from the expositions in this section so far. Provided the sharp drop in earnings does not occur too suddenly — i. providing the necessary repayments are still acceptable to the company (see subsection 9. a maximum debt rate close to zero in relation to liquidation value. on the basis of the expositions in this section. supervision of the company’s market value and of its volatility and of its debt rate . The situation is quite different if a sharp drop in earnings occurs very suddenly. a business policy decision. Here we intend to examine the resulting consequences. however — as outlined — dangerous to approach this maximum too closely. A sharp drop in earnings for any company that has been safely ﬁnanced as described above.7. the following tasks arise for the bank in the context of loan supervision: r supervision of the free cash ﬂows of the company seeking loans r derived therefrom. It is.7. Just how far a bank wishes to go. The company is then no longer in a position to make the repayments that would be required to reduce the debt rate to the necessary level concerned.7. then the result is logically.2) — there is no risk for the bank.5).3 Maximum Debt It has become clear from the expositions made in this section so far that no company should be granted more loans than are appropriate to the ﬁnancing of it for maximum return on equity. 9. at the end of the day. If it is pursuing the lending policy described above logically. If the company’s revenues fall so far that its market value becomes identical to its liquidation value.e.4 Consequences for Companies with Declining Earnings Any bank in the business of lending money must be prepared for companies seeking loans having to accept sharp drops in earnings. means — logically — that the following scenarios have occurred: r reduction in the company’s market value r increase in debt rate r possible increase in the volatility of the market value r increase in the risk-adjusted loan interest rate r the debt rate having been drawn close to that at which return on equity is maximised or having exceeded it (see Figure 9. in order to attain the required safety distance from the maximum debt rate (see Figure 9. The economic environment does of course play a major part in this too.7.5). 9. Thus there arises a case for the bank to wind up. and with what levels of repayment — and how quickly — it wishes to lead its borrower back from this danger point is.114 Risk-adjusted Lending Conditions 9. Thus the bank has no problem. it is in a position to make ongoing adjustments to the debt rate that are appropriate to the situation.

as deﬁned as a matter of business policy (see Figure 9.32) Companies with a debt rate that is higher than that of the maximum return on assets are in very serious danger of going bankrupt.7). In order to avoid cases of bankruptcy where there are sharp drops in earnings as outlined (Section 9.7. . particularly in the case of the ﬁnancing of far-reaching growth. If it does not do so. then bankruptcy is as good as inevitable. in order to protect the bank from damage. This therefore is also the maximum debt rate at which the bank should still be providing ﬁnance. if the sharp drop in revenues that befalls the company seeking the loan comes so rapidly that the repayment requirements arising necessarily from that are no longer acceptable to the company.20) The maximum achievable return on equity for a company corresponds to the marginal interest rate described above.8 RESULTS AND CONCLUSIONS The marginal interest rate of a loan plays a central part in the calculation of the ﬁnancial structures that are most favourable in terms of interest costs. It must be any bank’s objective that winding-up cases only arise. e = im (9. where the debt structure is appropriate. the bank must consistently call in the necessary repayments.Combination of Loans 115 r regular calculation of the maximum debt rate appropriate to the maximum return on equity r checking whether repayments have to be demanded in order to get back to the safety interval r laying down fresh risk-adjusted loan conditions r handing over care of the company seeking the loan to the bank’s winding up department. and in the determination of the maximum debt rate: ∂ i(L) (9. As long as a company has no more outside capital than matches the debt rate for maximum return on equity. It is. in order to be able to prevent bankruptcy in the case of sharp falls in earnings occurring. The maximum acceptable repayments correspond to the return on equity after taxation. The maximum debt rate.4) i m = i(L) + L · ∂L Several loans to the same company are structured most favourably in terms of interest costs if all marginal interest rates are identical: i mk = i mn for all k with 1 ≤ k ≤ n − 1 (9. as explained. recommended not to go so far in the provision of outside ﬁnance that this situation is reached. however. away from the maximum debt rate. the interest is still acceptable.4). In all other cases the necessary repayments should be called in consistently.5) 9. if required. to the extent that the company may no longer be able to make the necessary repayments. is thus also bound up with the question of whether the return on equity repayments permits the possibility of the debt rate returning ‘sufﬁciently rapidly’ to a ‘bankruptcy-resistant’ debt rate (see subsection 9.

38 7.44 0.11 0.96 5.01 5.04 0.23 5.74 10.42).75 4.41 0.43 48.03 4.49).62 48.12. The rate of interest i is calculated according to equation (4.49 0.6 again amounts to 4. however. as one would expect.67 4.86 13. if House A is mortgaged at 460 and House B at 540.2 Marginal interest rates House A mortgage ρ (%) 430 440 450 460 470 480 490 3.38).53 0.80 5.94 0. whereby the risk-free rate of interest according to Table 7.19 8.6 as in the example in Section 8.64 0.24 6.32 4. rounding differences may appear.37)–(7. This now leads to the ﬁnancing shown in Table 9.40 0.90 8.9 EXAMPLE The example in subsection 7.5).04 .25 5.07 0.26 4.38 4.65 4. The credit risks ρ ∗ ∗ result from the application of equations (7.78 ∗ ∗ ρ B ∩ C (%) ∗ House B ∗ i(%) i m (%) mortgage ρ (%) ρ B ∩ C (%) i(%) i m (%) Total interest 0.6 by a mortgage on a building.11 4. The marginal interest rate is calculated according to equation (9.5%.1113 according to Table 7. not as in the example in Section 8. The risks ρ B ∩ C are again the product ∗ of ρ times 0. times the loan interest rate concerned. in order to achieve optimum ﬁnancing.23 0. but the marginal interest rates.69 0.2.16 0.09 1. (Please note: if the ﬁgures in the following table are checked. At these ﬁgures the difference between the two marginal interest rates is also.6.11 5.07 48.55 5.06 5.91 7. whereby only that part of the table that is interesting for the purposes of the example is shown in Table 9.75 49.35 0.98 4.14 7.3. now be guaranteed.57 6.49 570 560 550 540 530 520 510 3.61 7.1 Starting out position Market value House A House B 800 600 Volatility of market value 50% 10% Table 9. This results in the following values for the interest curve. It is intended that the loan of 1000 over three years should.25 0.63 7. The difference between the two rounded up rates of mortgage interest amounts to 5 %.17 5. Table 9. at its lowest.37 4. on the most favourable conditions in terms of interest costs. The ﬁgures were originally calculated using an Excel worksheet to an accuracy of several more decimal points.1.29 6. The total interest results from the sum of the two mortgage amounts per line. This 8 shows clearly that not just the rates of interest.73 6.66 8. must agree with each other.27 49.) The total incidence of interest is at its smallest (48.53 5.11 2.58 0. The two properties have the characteristics shown in Table 9. but by two mortgages on two different properties.1 may again serve as the basis.62 0.42 48.116 Risk-adjusted Lending Conditions 9.54 48.

11% A 0.5% 5. . is connected with the much lower volatility of the market value of House B in comparison with House A. although its market value is signiﬁcantly lower.697% 4 11/16% 117 That House B in this example — in the case of ﬁnance that is at its most favourable in terms of interest costs — is mortgaged at a much higher rate than House A.1) Risk-free rate of interest Mortgage interest rate Mortgage interest rate rounded up 460 0.5% 4.58% BB 0.Combination of Loans Table 9.171% 4. The ﬁgures for this example were deliberately chosen to obtain results that would be as illustrative as possible.757% 4.297% 5 5/16% House B 540 0.3 Final results House A Mortgage Risk ρ ∗ Rating Rating risk (Table 2.

.

Part IV Implementation in Practice Procedure — according to the model — for assessing the risk in lending to a company Applications Final considerations .

.

So the answer to the question of how many sets of accounts from the past should be included in the assessment is far from trivial.1. The discount rate results from the CAPM. in order to take account of inﬂuences that may come to bear on volatility in the future.4. in order to avoid bringing into the assessment any factors that are now no longer relevant’. the value of the assets may never be smaller than their liquidation value. which must in turn support a forecast of the future asset values. in order to be able to ﬁnd out about the necessary model parameters as reliably as possible. 10. The volatility of the assets results from analysis of a range of annual accounts and budgets. on the basis of the discounted future free cash ﬂows. according to subsection 7. Another answer is: ‘only so far as the company’s past operational situation lines up with its current operational situation.10 Procedure — According to the Model — For Assessing the Risk in Lending to a Company In the normal course of events representatives of the lending bank discuss their business with representatives of the loan receiving company at least once per annum. The free cash ﬂows achieved in the past provide a clue for these. As was explained further in subsection 7.37) and (7. It follows from this that there must be a track record in terms of annual accounts for there to be any successful loan analysis. One answer to this question is — ‘as far as possible.1. The debt rate is worked out as the quotient of the outside capital divided by the value of the assets. with the aid of annual and/or intermediate accounts.38): debt rate.1 OVERALL VIEW OF THE PROCEDURE The following values have to be ascertained using equations (7. An important question here is how far back into the past to delve. in order to obtain the broadest possible statistical basis’. The value of the outside capital is taken from the balance sheet. six-monthly or quarterly accounts. and does in fact have to be answered afresh on each occasion. A check must therefore be made as to which of the two values is the greater (discounted free cash ﬂows or liquidation value). . Budgets should be brought into the calculations too. it being unimportant whether they are annual. We will demonstrate in this chapter how to proceed in this. volatility of assets and term of loan. The following documentation must therefore be brought together for the loan assessment: r A complete set of the company’s past accounts. as long as they have been drawn up according to the same principles.4. The value of the assets is determined.

In portraying sales in earnings statements it is often necessary to establish that distinctions are made between operational and non-operational. Here it must be ensured that the operating costs do not contain any depreciation or interest on debt. Here more value should be ascribed to reliability than to any r r far sight into the future. So all the company’s revenues must be added together when it comes to determining revenue. of determining sales. Other ﬁgures in earnings statements are irrelevant in connection with analysing loan risks. 10. When discussing accounts it should be ensured that the complete sales and operating costs for all existing accounts and budgets can be found out as precisely as possible. for determining the discount rate: return on the market. according to the model. to exist at least for the current and for the following ﬁnancial year. as the result is otherwise distorted. S. Clear ideas ought. The necessary ﬁgures.3 ANALYSIS OF CASH FLOW STATEMENTS The amount of investments is still missing for determining free cash ﬂow. . The same applies for costs as explained above in respect of revenues. as they do not feature in the calculation of free cash ﬂow. these then become an element in the company’s activity and therefore form part of its risk proﬁle. The remaining terms and notice periods of loans that have already been granted. yield of a risk-free investment in government bonds. This is worked out from the cash ﬂow statement.2 ANALYSIS OF EARNINGS STATEMENTS As mentioned in the preceding section. According to Brealey and Myers [BRMY96.122 Risk-adjusted Lending Conditions r Reliable budgets for the future. to be determined. provisions and interest on outside capital must be taken into account. On the other hand we will go. 71/72] the following applies: Free cash ﬂow = revenues − costs − investments The task is thus set. as it is the value of the company’s assets that have not been mortgaged that has. In the following sections we will on the one hand explain how this documentation should be analysed. however. Such distinctions are of no importance at this point. and between ordinary and extraordinary revenues. according to CAPM. and here all investments must be taken into consideration in turn. In particular depreciation and provisions are of no interest in this context. Net positions should be subdivided again into their individual elements. in each case. If not. Once the company’s management has decided that the enterprise should embark on certain activities. operating costs and investments. the determination of free cash ﬂow is involved in the analysis of earnings statements. one is entitled to raise the question of whether the company’s management has sufﬁcient specialist capabilities for the company to be credit-worthy at all. and the extent to which the non-mortgaged assets are at risk in the marketplace. All costs with the exception of depreciation. 10. into what values are critical for the analysis and which must therefore be probed appropriately in discussion with the company’s representatives.

for the discounting of the free cash ﬂows. In order to evaluate privileged claims correctly. 10. All the ﬁgures necessary for calculating the free cash ﬂows are thus determined. subordinated debts such as equity must therefore be considered. in order to be able to determine the current and future a debt rate. 10. It follows from this that in respect of the assets only their possible liquidation value in the event of bankruptcy should be determined. Here only debts in the category having precedence or in the same category have to be taken into account. no possible distinctions should be preferred according to the various categories. whichever value is the larger. Should a company have only quoted shares and .4 ANALYSIS OF BALANCE SHEETS Under the model. the discount rate applicable at that moment must be determined for each set of accounts. it may be assumed that these are correctly stated. If they are not. It follows from this that the calculation of the debt rate must be undertaken separately for each loan. When discussing accounts it should therefore be ensured that the company’s previous and planned investment activity can be found about comprehensively. The relevant β can be calculated as follows in the case of companies quoted on the stock exchanges [BRMY96. are of any signiﬁcance for the analysis of credit risk. if need be according to which category it falls into. Or. nor the company’s depreciation and provision policy. The determination of the risk measurement β plays a decisive role here.4. as already explained in the preceding section. S. Here any possible categories of loans must be taken into account.1) βV = risk measurement of the value of the company β D = risk measurement of the debts β E = risk measurement of the equity It may be assumed here that the values for β E and β D are available from suppliers of information such as Bloomberg or Reuters. 215]: βV = D E · βD + · βE V V (10. then it could be a case of falsiﬁcation of documents. If the debts are proved. When discussing accounts it should therefore be ensured that the liquidation value of the assets can be found out about as precisely as possible. On the liabilities side of the balance sheet it is only the current and planned debts budgeted vis-` -vis third parties that are of interest. the value of a company’s assets are worked out either from the value of the discounted free cash ﬂows or from the liquidation value of the assets. How the CAPM may be used for determining the discount rate was explained in summary in subsection 7.1.5 DETERMINATION OF THE DISCOUNT RATE In order to determine the respective value of assets.Assessing the Risk in Lending 123 And here too. put another way — neither the current operational value of the assets.

The determination of β is more difﬁcult in the case of companies that are not quoted. the periods of notice. The higher β-values should be used in comparisons. the shortfall risk of each of the company’s loans can be determined as described in Section 7. In the case of all other loans. however.4. the duration of the proceedings in any possible bankruptcy must be taken into account. by empirical testing. for the discount rates and for the liquidation values. at the end of the day. it is a question of the residual term until maturity.7 DETERMINATION OF VOLATILITY The volatility of the company’s value may be calculated. This question too may only be answered. Chapter 9: Capital Budgeting and Risk]). with the aid of three examples. Otherwise we refer. It follows from this that the volatility that is dependent on the term of the loan. which must be inserted into the model: r In the case of ﬁxed-rate loans.3. the term relevant to the credit risk corresponds to the sum of the time until the next discussion of accounts. In the case of ﬁxedrate loans. then the value for β D should be supported by values of companies in the same or related sectors.2. . 10. Here it must. and thus the credit risk. under the principle of conservatism. according to subsection 7. 10. the notice period is deﬁnitive. be ensured that critical situations are normally recognised only following a periodic discussion of accounts. the term relevant to the credit risk corresponds to the residual r In the case of loans on which notice can be given and of no ﬁxed duration. term plus the possible duration of any bankruptcy proceedings. Foreign values can possibly be used by way of comparison too. The method will be illustrated in Chapter 11.6 DETERMINATION OF RELEVANT LOAN TERMS When determining the relevant term for the loan using the model. Such comparisons are admittedly not simple in Switzerland. as described above. One solution here is to use the calculation of the β-values of quoted companies in the same or related sectors. as only a few companies are quoted compared with larger countries. So the following terms ensue. at this juncture.124 Risk-adjusted Lending Conditions not bond loans as well.8 DETERMINATION OF CREDIT RISK As soon as all values have been calculated as described in the preceding sections. The relevant loan terms result primarily from the loan agreements. to appropriate reading (for example [BRMY96. the question comes up of how long will it be until the bank granting the loan gets its money back. plus the possible duration of any bankruptcy proceedings. may be lowered to the extent that the terms of ﬁxed-interest loans. Moreover. or the intervals between discussions on accounts are kept short. plus the notice period. 10. on the basis of the values calculated — per set of accounts and per budget — for the free cash ﬂows.

00% 0 5 10 Mortgage 75%/Volatility 10% 15 Figure 10. S. It is furthermore recommended that new loans are checked more frequently at the beginning of their terms.1 in such a way that the resulting curve is about the same as in [FRRM97. empirically. such that ﬁgures based on pessimistic expectations are inserted into the model.00% 1. That is to say loans which can already point to some duration are much safer for the bank granting them than newly approved loans.00% 0.) It is therefore advisable to have particular caution prevail in the case of new business transactions. (The values for the mortgage and volatility were selected for Figure 10. The recognised rating agencies behave in just this way when it comes to assessment of asset-backed and mortgage-backed securities. too. not only from the bank personnel concerned.00% 3. in that they are tending to have to furnish more. This fact can be understood from the model described here. In these cases the rating agencies demand smaller loan reinforcements for the achievement of a top rating than for portfolios with newly granted loans.Assessing the Risk in Lending Probability of bankruptcy in the respective year of the term of the loan 125 4. for example.1 10. that default on loans occurs much more frequently at the beginning of the term than after a certain duration (see.9 PRUDENCE IN THE CASE OF NEW LOANS/BORROWERS It has been possible to observe. [FRRM97.10 POSSIBLE CAUSES OF CONFLICT BETWEEN BANK AND BORROWER WHEN THE MODEL IS APPLIED The application of the model described demands a fair amount of specialised knowledge and experience. S. Demands are made on borrowers. It is an advantage for a bank to put together for securitisation portfolios of loans in which the loans already have higher seasoning.00% 2. 112]). 10. and more detailed information about . 112].

combined with the careful examination of credit-worthiness required for that. which entails matching extra costs. This will be all the more difﬁcult. Such operational knowledge may. however. even if that is not the case. r The high loan interest rates will seem to threaten commercial viability. is.137 and thereafter]. not be presupposed to exist in all small and mediumsized enterprises. Particularly for the two reasons outlined above. It follows from this that auditors commissioned to produce annual accounts have to receive broadened mandates. the response on the bank’s part to any company having a high debt rate must be the application of appropriately higher loan interest rates. on top of willingness to share.6 and illustrated in Figure 9. have clearly proved this [BRUN94. while other banks (still) do not apply loan conditions that are consistent with risk. The commercial difﬁculties experienced by some banks at the beginning of the 1990s. these interest rates are also acceptable to the borrower and still make operational sense because it may increase its return on equity without running irresponsible risks. The application of risk-adjusted loan conditions. r The additional costs for auditors will be viewed as pointless expenditure. by means of detailed explanations. As was explained in Section 9. Such high-risk premiums on loan interest rates — as calculated by the model in some circumstances — have. application of the model may lead to conﬂict between bank and borrowers: r The scope of the information required will be seen as an unreasonable demand. been unusual in Swiss banking practice hitherto. . imperative for the continued existence of any bank. S. If the rules described in Chapter 9 are kept to. These causes of conﬂict should be countered by careful training of the bank ofﬁcials concerned and where applicable the borrowers.126 Risk-adjusted Lending Conditions their companies. Readiness to pay such interest rates if appropriate still has to be stimulated. however. This requires appropriate knowledge.5. however. and to insufﬁciently careful checks on credit-worthiness. and take all the longer. It does of course go without saying that this is far from simple. not least owing to far-reaching ﬁnancing of property and the excessively low mortgage interest rates that were associated with it.

so that the corrected loan risks are calculated (see Section 7. The proportions that the ﬁgures in any example bear to each other do. Loans are in each case granted for one year.3 may checking come up with some minor discrepancies.1. First the EBIT for the individual years is calculated and then capitalised using the long-term market return of 7%. while the results were calculated using the original ﬁgures. and on the strength of their past reliability. thus making it impossible to draw conclusions about the identity of the bank customers involved and maintaining bank conﬁdentiality. The Excel worksheet in Appendix 2 is used for the calculations. as the change in the starting values led to smaller changes in the proportions. Only in Section 11. which is fully understandable on the basis of the above analysis. assessed as competent. 11. whereby it is assumed implicitly that a value of β = 1 is. applicable in this case. so these were used for the analysis. Volatility is worked out using equation (7. on the strength of their previous experience. match. All ﬁgures we have given are not the actual sums in Swiss francs that were concerned. and the results are summarised in Table 11. with shops throughout Switzerland. So those responsible for the client .9): the ﬁgures below are obtained.45) on the basis of the company’s market value. however. In recent years this customer was classiﬁed by the bank as an increased credit risk. if one inserts 4% for the standard rate of interest. Then the company’s liquidation value for the individual years is determined from its assets. in order to make for accurate analysis. in the hope of better times.11 Applications In this chapter. in a turn-around situation. and in 1995 an external ‘turn-around’ adviser was brought in as well. Those in the bank responsible for the customer considered the company’s budgets very reliable. using normal banking correction factors. the turn-around expected for 1999 seems actually to be in sight. ﬁrst of all.1 SPECIALIST CLOTHING BUSINESS: TURN-AROUND SITUATION This example is about an established business specialising in ladies’ and gentlemen’s clothing. and budgets for the current year and for the next two years ahead. Annual accounts for the last ﬁve years. we will investigate three practical cases and compare the results with the conventional assessment made by the lending bank concerned. The loan was therefore continued. took over responsibility for the company at the beginning of the 1990s. An approximate ﬁgure of 20% of annual wages and salary costs is inserted for privileged salary claims in the event of bankruptcy. according to the CAPM. A new management team. The higher of earning capacity value and liquidation value is taken as the market value for further calculations. were available for analysis. The applicable debt rate is calculated with reference to outside capital owed to third parties. The budgets for 1999 to 2001 are encouraging.

6 Rating according to Table 2. When it comes to setting the interest rate here. An EBIT for 1999 of 100 is budgeted for distribution. The advantage of the model consists.1 Privileged salaries 56 58 Volatility of the market value: calculated: 17. An average rate of interest on debt of 10% is realistic for the bank in the context of the budget being attained. the positive prospects should at least be conﬁrmed by means of a set of intermediate accounts as at 30.0%∗ Annual credit shortfall risk (in %) 24. being of no further interest by spring 1999. The uncertainties regarding the quality of budgeting. because of the strained credit-worthiness situation. however. Equity as per the balance sheet of 31.1 60 EBIT 70 84 66 76 74 Earning rate 1000 1200 943 1086 1057 Breakup value 1060 960 920 800 820 Market value 1060 1200 943 1086 1057 Debts 1076 1048 Debt rate (in %) 99. The positive prospects for the future do.3 Corrected annual credit shortfall risk (in %) 49.1 99. NB: The ﬁgures for 1994 to 1996 were not calculated out.12 0. Before anything is undertaken on this.6. however. this may neither be based on the current situation (too pessimistic) nor on the budget projection (currently as yet too optimistic). allow their continuation to appear appropriate. of which 5 remain as earnings and 95 are set aside for debt servicing. and the capabilities of the management team were taken into account with . The earnings reported for 1999 should therefore be split into 100 parts.1 Specialist clothing business: model information and results Five annual accounts Year 94 95 96 97 98 99 100 1429 820 1429 1000 70. The bank-related debt amounts to 938. it seems premature.99. the remaining debt being non-interest-bearing positions such as creditors. in the fact that the risks involved are quantiﬁable and appropriate loan supervision measures may be laid down. anyway.37 0. rather. that loan risk assessment must be underpinned by parameters set in the future because — on the basis of the current situation — the loans to this company would have had to be terminated without notice.22 BBB 0.2 60 01 122 1743 800 1743 908 52.5 24. This example allows us to show.128 Risk-adjusted Lending Conditions Table 11.7% rounded: 20. But in order not to jeopardise the turn-around by imposing excessively high interest costs. at least a part of the loans might be converted into a proﬁt participating loan with returns thereon being dependent on success. impressively. Such intermediate accounts have been being obtained for a long time.0 60 Three budgets 00 106 1514 820 1514 972 64. for the interest rate to be set on the basis of what the company can accept to pay. The conventional loan decision and the loan decision emerging from the use of the model are qualitatively identical in this example.1 DD DD 0.009 AAA ∗ Volatility was rounded up to 20% to take uncertainties into account. have applied to the loan assessment ofﬁce for the client to be assigned to a higher category of credit-worthiness within the bank.66 BB 0.12. the development of the market and of the sector. As this application is only supported by budgets. This combination of circumstances calls.004 0.0 49.98 comes to 82.

64 0. based on the 1997 annual accounts. Analysis of the ﬁgures made available has resulted in the company’s market value in all years having been higher than its liquidation values. on the basis of a freshly drawn up business plan. He intends.007 0.7 CCC 192 2.5 1. this dividend income corresponds to the company’s EBIT.66 BB 144 2.1 Rating according to Table 2.0 0.1 1. and offers the younger his shares at a price of 240. being of no further interest by spring 1998.0% loan 240 ∗∗ Duration loan 3. The three most recent sets of annual accounts and ﬁve forward budgets of the trading company are available for this examination. The loan application has therefore to be examined on this basis.0 AAA of the dividends of the trading company. This leads in 1998 to a debt Table 11. An average loan risk of 5.5% of its share capital.0 0. Disregarding minor management costs.0 5.5 0. The relevant ﬁgures are shown in Table 11.0 Credit shortfall risk (t) (in %) 14.8% rounded 60.2 COMPANY TRADING IN MACHINE TOOLS: PROVISION FOR SUCCESSOR COMPANY The two major shareholders in a company trading in machine tools each own 48. The older of the two major shareholders would like to withdraw from the business. .1 CC ∗ Corresponding to 97% ∗∗ Disregarding interest.005 AAA 48 1. with payments of 48 per annum. Only the 97% of the future dividends of the trading company will be available to service the debt. whereby he would pledge the entire shareholding of 97% in his ownership in favour of the loan.2 (calculations with the help of the worksheet in Appendix 2).28 0. this makes a loan interest of about 10%. EBIT will again be capitalised at 7% in order to determine the company’s market value. The latter intends to form a holding company to acquire the shares. together with a third person who owns the remaining 3% of the shares. They form the management team.7 Average annual credit shortfall risk (in %) 5.Applications 129 an increase in volatility. 11. NB: The ﬁgures for 1995 and 1996 were not calculated out. The liquidation value of the company matches the current value of the shares.2 Trading company: model information and results 3 Annual accounts Year ∗ ∗ Five budgets 97 98 93 1329 99 97 1386 00 98 1400 01 101 1443 02 98 1400 95 96 Holding-EBIT 15 13 52 Market value 214 186 743 Market value volatility 56.0 0.14 A 96 1. 240 3. The banker’s experience plays an important part in this.1% results. With a standard rate of interest of 4%. Here it is once again shown that establishing the ﬁgure for volatility is of decisive importance. to repay the loan over ﬁve years. He applies to the bank seeking a loan for the full purchase price.

This is acceptable under the 1998 budget.1 Starting Position A major shipowner has had a group of six bulk tankers ﬁnanced by a consortium of eight banks. deliver detailed reasoning for having to turn down the loan application. according to estimates by two internationally respected ﬁrms of shipbrokers. Cash ﬂow analysis of future revenues worked out that ships’ incomes of about 0. It is apparent from this example that budgets that are very optimistic compared with past performances do lead to high volatility in a company’s market value. The business was not done.3. this means that any fast growth must be ﬁnanced above all by its own cash ﬂow. The model does.3. considerable loan risk would have ensued from minor mortgaging of just about one-third (240/743 = 0. The banks granted the loan application on the strength of the low proportion of the ships’ value mortgaged and of the fact that it had been possible to extend their working lives through to the middle of 2003. The budgets do.98. in turn. was put at a total of 1000 at 29. but not in relation to the 1997 accounts.3 SHIP MORTGAGES: RISK LIMITATION This example is all about ﬁguring out where the risk in the loan transaction lies. viewed in relation to the past. moreover. The ships were built in 1976/77. conversely. The qualitative loan decision reached both by conventional means and via the model was identical in this example too. The questionable acceptability was also decisive. That is therefore no solution.329). Due note was taken that the loan outstanding of 171 as at 30.11 per day at sea would sufﬁce to cover the costs of operating and maintaining them.130 Risk-adjusted Lending Conditions service of 24 (interest) + 48 (repayment) = 72. thanks to the reﬁts. because of the high volatility associated with it: ﬁnance using bank loans becomes rapidly associated with high risks that lead. The shipowner applied in the middle of 1998 to increase the loans to 450. Related to any operating company.6. 11.98.3. Any reduction in the repayment would. 11. 11.0. For this loan application the bank had also established by conventional methods that it should refrain from granting the loan. to matching high (too high!) interest costs. The budgets were assessed as being too optimistic when it was taken into account that an experienced member of the management team was leaving the company. in order on the one hand to ﬁnance reﬁtting costs and on the other to have liquid funds available for the purchase of new ships.2 The Banks’ Loan Decision The banks consented to the increase in the loans and agreed with the shipowner the repayment plan in Table 11.9. As a result of costly special surveys the value of the ships. Ship mortgages have been set up for this purpose. The loan application should therefore be declined on the grounds of questionable acceptability.03 would . lead to an extension in the duration and thus increase the loan risk and interest rate. and the loan costs. however. seem really rather optimistic. The loans were 378 at 30. In this case. The scrap metal value of the ships at the same point in time amounted to 212.

**Applications Table 11.3 Ship mortgages: model information
**

From end of quarter Amortisation Loan Depreciation III 98 IV 98 I 99 II 99 III 99 IV 99 I 00 II 00 III 00 IV 00 I 01 II 01 III 01 IV 01 I 02 II 02 III 02 IV 02 I 03 II 03 13.1 15.2 19.8 13.6 13.6 13.6 18.0 13.6 13.6 13.6 18.0 13.6 13.6 13.6 18.1 13.6 13.6 13.6 13.5 450 436.9 421.7 401.9 388.6 374.7 361.1 343.1 329.5 315.9 302.3 284.3 270.7 257.1 243.5 225.4 211.8 198.2 184.6 171 41.5 41.5 41.5 41.5 41.5 41.5 41.5 41.5 41.5 41.5 41.5 41.5 41.5 41.5 41.5 41.5 41.5 41.5 41.0 Ship value 1000 958.5 917.0 875.5 834.0 792.5 751.0 709.5 668.0 626.5 585.0 543.5 502.0 460.5 419.0 377.5 336.0 294.5 253.0 212 Debt rate (in %) 45.0 45.6 46.0 45.9 46.6 47.3 48.1 48.4 49.3 50.4 51.7 52.3 53.9 55.8 58.1 59.7 63.0 67.3 73.0 80.7

131

High Low volatility (in %) volatility (in %) 43.0 41.8 40.6 39.4 38.2 36.9 35.7 34.5 33.3 32.1 60.9 29.7 28.5 27.3 26.1 24.8 23.6 22.4 21.2 20.0 27.0 26.2 25.4 24.6 23.8 23.1 22.3 21.5 20.7 19.9 19.1 18.3 17.5 16.7 15.9 15.2 14.4 13.6 12.8 12.0

certainly have been covered by the scrap metal value of the ships. The positive decision was above all underpinned by the following considerations:

r At 45%, the proportion of value mortgaged was small at the time of the increase in the loans. r The loan repayment at the end of the ships’ lives — the ‘balloon’ payment — is well covered r According to the bank’s calculations the break-even freight rate amounts to about 0.09 per r

day, based on details supplied by the shipowner and on comparison with similar ships — the current freight rate of 0.11 per day was thus quite well sufﬁcient. The ship owner’s management team was judged to be very capable and reliable. by the scrap metal value of the ships that was expected at that time.

11.3.3 Assessment of the Loan Risk by the Banks The bank that made the documentation available had assessed the loans with a loan risk of 80 basic points according to its internal rating system. This classiﬁcation was made on the strength of comparisons with loans in similar positions. 11.3.4 Determination of Loan Risk According to the Model Additional factors for determining loan risks according to the model must include the volatility of ship prices and of scrap metal prices, realisation costs in the event of bankruptcy, time scales and ship depreciation:

132

Risk-adjusted Lending Conditions

Volatility Details from the spring 98 issue of the bi-annual Clarkson Shipping Review and Outlook — from the Clarkson ﬁrm of shipbrokers — were used for determining the volatility. Both the time series from 1979 to 1998 for the tankers in question (Page 16, column 280k 1975) and the relevant scrap metal prices (Page 153) are to be found there. The following volatility ﬁgures for each of the years concerned may be calculated by using this information: Ship prices: Scrap-metal prices: 1979–98: 1993–98: 1979–99: 1993–98: 43% 27% 20% 12%

Behind the use of two different time periods is the consideration that the 1979–98 sequence of numbers allows calculation of the worst case, and the ﬁve-year 1993–98 period is most likely to be representative for the last part of the term of the loans. As it was possible to depreciate the ships using the straight-line method (see below), it may be assumed for the sake of simplicity that volatility of the ship prices moves in line with volatility of the scrap-metal prices (see Table 11.3). Realisation Costs in the Event of Bankruptcy These amount to about 4.5 per ship, according to the bank’s information. Costs of about 27 have therefore to be allowed for six ships. Time-scale Considerations The loan agreement contains a clause covering the event of default. Thus, if the shipowner is at any point not in a position to service the debt on time, then the consortium of banks has the right to make the loan repayable immediately and to realise the value of the ships. According to the bank’s information, such ships can be realised within three to six months, be it privately or by means of auction. The loan may be utilised on a roll-over basis in the form of advances in Euro of, in each case, terms of from one to 12 months. The longer-term advances have intermediate interest deadlines, in each case after three months. So every three months the banks may establish whether the shipowner is still in a position to service its debts. On top of the time needed for realisation of the ships, a reaction time of from six to nine months is worked out for the banks. The following risks are examined in terms of their time-scale aspects on the basis of this situation:

r High volatility: 6, 9, 12 and 15 months from the most recent proper payment service the r Low volatility: 6, 9, 12 and 15 months from the most recent proper payment service the debt.

Depreciation of the Ships According to the bank’s information, the ships may be depreciated using the straight-line method from their current value of 1000 as at 29.9.98 to their scrap value of 212 as at 30.6.03. debt.

Applications Table 11.4 Ship mortgages: results in % Whole term From end of quarter III 98 IV 98 I 99 II 99 III 99 IV 99 I 00 II 00 III 00 IV 00 I 01 II 01 III 01 IV 01 I 02 II 02 III 02 IV 02 I 03 Low volatility 0.5070% Term High volatility 2.8073% Term

133

6 months 9 months 12 months 15 months 6 months 9 months 12 months 15 months 0.0004 0.0003 0.0002 0.0001 0.0001 0.0001 0.0001 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0001 0.0007 0.0095 0.0077 0.0063 0.0048 0.0030 0.0024 0.0021 0.0017 0.0011 0.0009 0.0008 0.0008 0.0005 0.0005 0.0007 0.0010 0.0011 0.0028 0.0102 0.0607 0.0318 0.0271 0.0219 0.0151 0.0127 0.0113 0.0097 0.0069 0.0060 0.0054 0.0051 0.0038 0.0039 0.0046 0.0062 0.0069 0.0137 0.0373 0.1487 0.0749 0.0653 0.0545 0.0401 0.0346 0.0314 0.0276 0.0207 0.0185 0.0169 0.0161 0.0124 0.0129 0.0146 0.0184 0.0202 0.0353 0.0802 0.2512 0.2025 0.1795 0.1530 0.1156 0.1022 0.0886 0.0797 0.0617 0.0568 0.0537 0.0527 0.0427 0.0460 0.0537 0.0699 0.0716 0.1266 0.2839 0.8509 0.6320 0.5751 0.5091 0.4142 0.3762 0.3369 0.3101 0.2569 0.2404 0.2294 0.2245 0.1927 0.2013 0.2230 0.2669 0.2705 0.4027 0.7136 1.5808 1.1287 1.0398 0.9373 0.7904 0.7273 0.6613 0.6148 0.5265 0.4961 0.4746 0.4634 0.4085 0.4200 0.4525 0.5158 0.5216 0.7119 1.1211 2.1343 1.6099 1.4939 1.3610 1.1724 1.0868 0.9968 0.9320 0.8135 0.7692 0.4368 0.7179 0.6423 0.6357 0.6921 0.7719 0.7721 0.9997 1.4655 2.5463

The model information needed according to the above expositions may be taken from Table 11.3. The results are summarised in Table 11.4. It was assumed, in the case of the calculations of the quarterly loan risks, that the situation as it was then is presented as in Table 11.3, and that the shipowner had up until then met its commitments fully on each occasion. The calculations were made with the aid of the worksheet in Appendix 2. As may be seen in Table 11.4, the loan risk is heavily dependent on the bank’s reaction time. The ﬁgures in the case of 15 months are signiﬁcantly higher than for six months, under the assumption both of the lower and of the higher rates of volatility. It becomes apparent that assessment of the volatility of the value of the collateral and the bank’s reaction time acquire decisive signiﬁcance in the case of this loan. It is more difﬁcult to frame the risk calculation over the full term. The assumption was ﬁrst of all made that servicing the debt had until then taken place in full at the beginning of any calendar quarter. It had therefore been worked out at ﬁrst, for calculating the loan risk, how big it would have been on the assumption that the remaining debt was still outstanding at that point in time. Then the same loan risk was calculated as at the end of the quarter, and the loan risk was determined for that quarter from the two ﬁgures taken together. The weighted average, in terms of amount, of these risks per quarter, converted to one year, then give the loan risk per annum over the entire term (Table 11.4).

for each individual quarter for three or six months (i. So it has to be assessed whether it is more advisable for the bank to realise a lower price in the short term. this is a worst-case scenario.e. It follows from the above reﬂections that the bank assessed this loan very cautiously. except in the last two quarters. but they are still very small. over a somewhat longer time-scale. justiﬁably estimated less pessimistically. As may be gathered from Table 11. loan risks that are still less than ﬁve basic points result from reaction times of 12 and 15 months (corresponding to realisation periods of 9 and 12 months). depending on the risk situation. however.e. It is only here that the decisive inﬂuence of the volatility ﬁgures used becomes apparent. ﬁgures of from 4 to 72 basic points emerge for the two shorter reaction times. with 25 basic points credit risk. a price that is higher but still uncertain. however. including the time expected for realisation) a normal credit shortfall risk of less than one basic point.5 Comparison of Assessment between the Bank and the Model There is a credit shortfall risk of about 51 basic points on the assumption of the lower rate of volatility. the bank’s assessment is more or less in accordance with the model’s results for higher volatility rates.4 show that the credit risk rises sharply if the time taken to realise the ships were to draw further out or if the credit risk is considered over the entire term of the loan. From the point of view of the individual quarters. When considered in this sort of way. and with the exception of the last quarter.) 11.4. Seen from the point in time at which the loan is discussed there results. The more optimistic evaluation of the credit-worthiness of this ship mortgage can thus only be justiﬁed if: . It follows from this that the bank assessment lines up more or less with the ﬁgures for the higher rate of volatility taken from the model. There is a credit shortfall risk of about 2. in line with its internal rating system.3. (The bank later decided on the more optimistic evaluation of the position. realisation of the ships may or may not be forced in terms of time. For reasons of conservatism it may be argued that a credit rating must be aligned to a longer than normal realisation period. On the other hand. because of the high inﬂuence of the bank’s reaction time on the loan risk in the event of default. But this too is again a worst-case scenario. 11.134 Risk-adjusted Lending Conditions A situation assessment has to be carried out ﬁrst of all.8% on the assumption of the higher rate of volatility. Only for the last two quarters are the ﬁgures correspondingly higher. considered over the whole term of the loan.6 Limitation of Loan Risk The results in Table 11. If volatility is. But as every three months it can be viewed again whether the shipowner is or is not still meeting its commitments. the current value of the ships and the volatility of the development in value may not be inﬂuenced. In the ﬁnal quarter the ﬁgures are again somewhat higher. because of the higher proportion of mortgage existing at that point.3. Only in the case of reaction times of one year or more would the risk have to be estimated as being higher. i. then the bank — without further ado — may classify the loan one level better. The amount of the loan. It follows that the bank assessment is too pessimistic on the assumption of the lower rate of volatility. considered over the full term of the loan. or whether it should attempt to achieve.

It may also be demonstrated what risks the bank runs if the necessary loan supervision measures are not undertaken. Figure 11. does on the one hand show where the credit risks can be limited. of from three to six months. Our intention now is to examine whether this situation could have been foreseen with the aid of the model. The property index drawn up by the Cantonal Bank of Z¨ rich provides u the basis for the calculations (see Appendix 3). 11.Applications 135 r The credit-worthiness can be reassessed every three months. realistic. It does. It makes it clear that the index is at its most volatile for multiple dwelling units. The above listing of preconditions. and if necessary the default r The short time-scale for realising the ships. thanks to the model. and of the relevant inﬂuencing factors involved. is considered to be r The volatility rate for the value of the ships is assessed. The model’s informative value is therefore substantially higher than conventional methods of loan assessment. however.4 MORTGAGE BUSINESS 1985–99 Over the period 1990–95 the banks in Switzerland were confronted with high losses on mortgage business. The mortgage business is 300 250 Index 200 150 100 50 1980 STE MFH EFH 1985 1990 Year 1995 2000 MFH is the abbreviation of the German text for multiple dwelling unit. STE is the abbreviation of the German text for condominium. This example shows plainly that. Figure 11. also become clear that a bank requires much experience in the ﬁeld of ﬁnancing international shipping to be able to operate this sort of business proﬁtably. clause can be invoked. in line with the residual term of the loan — on the optimistic side. on the basis of a ﬁve-year time-scale – r The shipowner’s management team is considered capable of operating the ships over the coming ﬁve years in line with revenue expectations.1 gives a graphical representation of the course of the index. the credit risk may both be quantiﬁed and also limited.1 . EFH is the abbreviation of the German text for single-family home.

e.2 21. This was calculated for each calendar year with the help of the ﬁgures for the last four. It is striking that the volatility ﬁgures show a clear peak at the beginning and at the end of the 1990s. which leads to different mortgage ﬁgures in each year corresponding to the course of the index. It is of course debatable whether this is the right way to proceed. ﬁve and six year’s indices.2 15. The course of the volatility is detailed in Table 11. Two assumptions vis-` -vis mortgages are made for the credit shortfall risk calculation.3 24.2 7.2 20.5 Volatility of the index in % — for multiple dwelling units as per Appendix 3 Year 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 4 Index values 10.6 21.2 17. the volatility in each case per annum is determined in this way.5 7.7 8.2 25.136 Risk-adjusted Lending Conditions Table 11.7 1.0 15.4 5 Index values 8.5 19.45)).5 7. such that three. four and ﬁve quotients respectively were available for the calculation (see equation (7.4 7.3 6. The highest volatility ﬁgure concerned is given in each line in the last column of Table 11.4 7.2.2 19.8 10.6 18.9 7. in line with the assumption that neither repayments of them.3 6.6 17.9 15.4 9.8 19.1 18. The credit shortfall risk ﬁgures calculated in this way demonstrate what sort of risk the bank was running in each calendar year concerned by ﬁnancing to the extent of 80%.6 8.2 7.2 20.2 Maximum 10. The credit shortfall risk ﬁgures calculated in this way demonstrate how they developed for the bank over the years.3 16.2 7. For the purposes of this example.and long-term volatility are included in the later calculations.7 6. is mortgaged at 80% of its saleable value. however.4 19. The a ﬁrst assumption is that a multiple dwelling unit in 1985.2 21. This means that both the inﬂuences both of short. i. The second assumption is that in each calendar year in each case a multiple dwelling unit is mortgaged again to the extent of 80% of its saleable value. .3 23.7 2.2 3.5 8.4 therefore investigated. well before the mortgage loan crises at the beginning of the 1990s.9 5.4 18. The cause is in the ﬁrst place the sharp increase in the index ﬁgures and the subsequent sudden transition to a decline in prices.8 17.5. without repayments.3 23. The amount of the loan is left unchanged in subsequent years. The ﬁgures in the table are shown graphically in Figure 11.7 7. in what follows.2 25.5.8 10.6 24.7 2.7 20.9 6. took place.7 6.5 3.4 9. and the uneven development in the second half of the 1990s. These ﬁgures are used later in the course of this analysis to deﬁne a worst-case scenario.5 9. using this index.4 6 Index values 7. nor increases in them.5 6.7 17.

based on the annual index ﬁgures.00 1988 0. Appropriately higher ﬁgures should be used when granting ﬁxed mortgages over several years. 2 1985 0.27 1998 1.00 1996 0.00 0.6.00 0. the credit shortfall risks under the ﬁrst assumption were very small even during the crisis at the beginning of the 1990s.07 1.00 0. Annual volatility ﬁgures were used for the calculation. Under the second assumption. in the immediate future. By using quarterly ﬁgures it may be assumed that the reaction time could have been reduced even further.2 As may be inferred from Table 11. As early as 1990 the ﬁgure was so high that 80% mortgages in that year were no longer appropriate.82 2000 137 30% 25% 20% Volatility 15% 10% 5% 0% 1985 4 5 6 Max 1990 Year 1995 2000 Figure 11. they became plainly higher than previously at the beginning of the 1990s. 1 Ann.99 2.10 3. however.08 1994 0. to track down whether the credit shortfall risks rates that have just risen again are leading to a second mortgage loan crisis in the near .03 1997 0.85 1990 0.09 1993 0. was foreseeable from the model as early as the beginning of 1991. It becomes clear from Table 11.09 0. It would be interesting. after the index ﬁgure for 1990 came into existence.6 Credit shortfall risk development in % under the assumptions Year Ann.00 0. 2 Year Ann.98 1986 0.00 0. The facts are portrayed graphically in Figure 11.28 1.91 1999 0.00 0.57 1992 0.Applications Table 11.6 and Figure 11.18 1989 0. This situation repeated itself towards the end of the 1990s.01 3. which leads to credit shortfall risks that are correspondingly even higher.3.90 1991 0. 1 Ann.01 1995 0.88 2.3 that the mortgage loan crisis.00 1987 0.02 1.00 0.

however. Such testing would. Whether or not the results are also sufﬁciently accurate in quantitative terms could only be established by comprehensive empirical testing.138 Risk-adjusted Lending Conditions 4% 3% 2% 1% 0% −1% 1985 1 Credit short fall risk 2 1990 Year 1995 2000 Figure 11. have been taken. on the basis of the experiences banks have had. or whether the necessary safety precautions. at least in qualitative terms. On the strength of this result the usefulness of the model proves itself.3 future. far exceed the scope of this study. .

introduction and applicability are connected with each other.3 what the preconditions are for its applicability. for example [BOHL92]) to determine whether deviations between forecast and actual losses are coincidental.12 Final Considerations So far we have emphasised that the method of calculating risk-adjusted loan interest rate presented here is only a theory. And here checks must be undertaken. yet necessarily mean that the results are also precise in quantitative terms. It follows from this that the empirical examination must be built up by analogy with preliminary and actual costing in the insurance industry.4. 12. using recognised statistical tests (cf. mathematical model for calculating the risks involved for banks in their lending business.6. In Section 12.8 too). As the examples in the separate chapters show. Deviations that are coincidental lead to the conclusion that the theory does hold up.2 and from the reading referred to there. So it has to be tested empirically.1 TESTS OF HYPOTHESES The method presented here is a theoretical. It was also pointed out that this empirical testing would exceed the scope of these expositions. Section 1. however. as may be inferred from subsection 7.8.1 how this testing should be undertaken. We will nonetheless outline in Section 12. In the cases of . and in Section 12. r that the parameters selected do not hold up. and in Section 12. So it is necessary. S. But this is none other than the comparison of imputed ‘insurance premiums’ with actual ‘loss experience’ (cf. We will conclude this study with some closing remarks in Section 12.2 how a bank wishing to introduce the method described here must proceed. We will explain in Section 12. or whether they follow a pattern. in the next step. Deviations that follow a pattern may have two causes: r that the theory does not hold up.8). the results of the method are plausible as they match up in qualitative terms with the experience of banks in this business. That does not.5 the questions still open will be summarised. whether the ex ante forecasts losses agree with the ex post actually established losses using the method presented here. Theories have to be examined empirically in order for their validity to be corroborated (see Section 1. 50:a] backtesting). We will at the same time point out how closely the empirical testing.4 possible customer considerations will be discussed. The construction of the empirical test is once again based on the concept of insurance. to test the method empirically (see Section 1. [BCBS99. It is above all the determination of the ‘correct’ rates of volatility that is difﬁcult.

S. Section 1. S.8 [BCBS99. r r r r .1)). 50: d)]). The losses actually incurred in each interest period. The precautionary provisions actually made for each interest period. equation (1. The operational expenditure (personnel costs. Section 1. Section 1. 50: b) and c)]). Even today there is probably not even one bank that is aware of how high are the costs of processing just one single case of winding up. with customers). which can be subdivided into three main groups: r Creating the preconditions for the application of the method described here.1) for each interest period. above all. possibly. r Creating the preconditions in operational accounting terms (also. Preconditions for the Application of the Method Described Here These will be elaborated in the next section.140 Risk-adjusted Lending Conditions deviations that follow a pattern it must therefore. indeed. but they have to be covered too and. These costs do not in fact have anything directly to do with the costs of risk discussed here. then this too may be foreseen at the same time. subheading ‘correlations’). be checked whether any better agreement between the forecast and the actual losses might have been achieved by selection of ‘better’ volatility rates. If the operational accounting is already in place. This becomes evident as one becomes aware of the prerequisites.2 on ‘statistical bases’. Then we must ﬁnd out how such ‘better’ volatility ﬁgures could have been determined ex ante. if required. r Creating the preconditions for the facts to be analysed. Preconditions in Operational Accounting Terms In order to obtain the widest possible factual basis for statistical tests which in turn allows for the most varied analyses. An independent examination of the method of procedure is also needed so that the model.5.2.) incurred in processing winding up operations in each interest period. The ﬁgures necessary for determining correlations in each interest period (see subsection 12. by proﬁt contributions p · L (cf. Only if this approach is unsuccessful should the theory be rejected (cf. either the risk premiums actually demanded of the customer are determined.8 [BCBS99. becomes acceptable for regulatory purposes (cf. or the risk premiums calculated according to the model on the basis of the theory are obtained and imputed. divided — for the purpose of further analyses — into those required for regulatory purposes and those required operationally for the banks’ own estimating purposes. it is necessary from the start to record the following facts for each loan utilised: r The agreed risk insurance premium r · L (see equation (1. etc. Depending on whether or not the method is already being applied in the test phase on a 1/1 basis. The expenditure required for the empirical testing outlined should not be underestimated in any way at all.

Final Considerations 141 Clearly the expenditure required on IT for this is considerable. at a later stage. The action described above will clearly be very expensive in time and money. It is also the case that a majority of bank loans consists of standard transactions such as advances against securities. but it is nonetheless essential for achieving our purposes. both in the ofﬁces that assess and wind up loans.5. the necessary IT resources have to be made available. Simpliﬁcations may be made. up to a certain point. Leading bank ofﬁcials. On top of that banks will have to recruit the necessary specialist staff who must be capable of designing and evaluating statistical tests. It is necessary to get right on top of the data ﬂow that the facts described above will form.1 Specialist Personnel The necessary specialist personnel must be recruited for successful implementation of this method. It must be assumed that focused management of such a heterogeneous team will be critical. ﬁnancing of owner-occupied houses. the granting of loans for large ranges of mortgage and volatility ratings is completely without problems. For this. as follows: r specialist personnel r statistical bases r IT support r organisational measures 12. if need be. and which have exceeded the dimensions they were expected to have at the time. Loan Assessment Ofﬁces The will must be there to recruit specialist personnel who have sufﬁcient ﬁnancial and mathematical knowledge to be able to understand this theory. The implementation of empirical testing must therefore be carefully planned. As becomes evident from the illustrations in Section 7.2 IMPLEMENTATION IN PRACTICE We can group the preconditions that must be fulﬁlled for the putting this method into practice. in the form of a project. But specialists who are capable of making their own calculations are needed for assessing more demanding loan transactions — whether it is more complex loan structures that are . The costs that are budgeted and the ongoing costs that will then arise have to be set against the losses on loans that have been incurred in the past. in turn. if the theory has proved well founded and when it has been possible to gather together the ﬁrst experiences of its application by converting unique data into composite data.2. and budgeted for. 12. and ﬁnancing of small companies on current account credit. It is at this juncture that we venture to forecast that this comparison will indeed demonstrate that the expenditure for such a project will be justiﬁed. It may be assumed that the great majority of such business can be processed by standard. IT-supported procedures. actuaries and economists familiar with the methods described here will have to work together in such specialist teams.

there are quite clear upper limits when it comes to the mortgaging of owner-occupied houses.4. They must most of all be able to demonstrate. Winding Up Ofﬁces The specialists in winding up ofﬁces must be capable. using illustrations. Financing Companies In subsection 7. i. The results in these cases in particular must be probed.142 Risk-adjusted Lending Conditions concerned. even if much can be made easier by IT support.4. for determining the volatility of the market value of a company. The borrower against securities is particularly concerned to prevent excess collateral for his loan being taken.2 Statistical Bases We have referred repeatedly to the importance of having available the necessary statistical bases. Some of this can most probably be standardised in terms of IT. On the other hand. and vice versa. the lower the rate of interest. of undertaking the calculations described in Section 9. in cases of loan transactions with simple structures. with the aid of tables and charts. Here it must be demonstrated. at this point.2) are particularly critical in this respect.2. or ﬁnancing arrangements that are more far-reaching. borrowers must be in a position to deliver all this information. that these preconditions are not sufﬁciently fulﬁlled in the case of small and mediumsized companies. above all. By way of contrast. their acceptability is no longer to be assumed.4. Sales Sales staff must be capable of explaining the loan conditions that have been calculated to customers. and subjected to sensitivity analysis. the correlations between mortgage.e. Similar considerations apply to the ﬁnancing of enterprises. volatility and loan interest. whether or not the volatility was ascertained in this way with sufﬁcient precision. while the owner-occupier is particularly interested in obtaining the highest possible mortgage on his house — especially when he is ﬁrst buying it. 12. These play an especially important part in advances against securities and the ﬁnancing of owner-occupied houses.6.2 we made the assumption that. Regretfully must we record. One way of improving this situation lies in the opportunity it provides for . where in the case of high mortgages interest rates rise sharply such that. There is thus no sense in which there can be excess collateral in the case of an advance against securities. beyond a certain point. Two questions arise here: on the one hand it must be checked empirically. some three to four sets of annual accounts and one or two budgets for the ﬁnancial years ahead must be available. the higher the collateral. A more profound understanding of this theory is needed for such cases. The ﬁgures for volatility (see subsection 7.2. that the level of mortgage in connection with volatility does have a direct inﬂuence on the loan interest level. for inaccuracies. in accordance with the expositions in subsection 7.

These would normally be software developers working on technical/scientiﬁc applications. Work in this ﬁeld must be carried further and extended to cover the whole of Switzerland. The resulting high loan interest might possibly motivate borrowers in future to prepare the necessary facts and ﬁgures. Due attention must be paid to the numerical methods used. because of the complexity of the formulae.2. No great volumes of facts actually have to be processed in the calculations relating to any one single loan. the correlations that will be of interest. Loans Against Portfolios of Securities The volatility of any portfolio of securities can be determined by reference to stock exchange statistics. in Section 12. in order to be able to calculate. Figure 8. On top of that. based on that. c. statistical bases and IT solutions. of staff recruitment and training. Correlations The correlation between the probabilities of borrowers defaulting and of collateral falling short was studied in Sections 8. The Cantonal Bank of Z¨ rich has u adopted a pioneering role here. .1.2. IT support for creating preconditions in terms of operational accounting and in terms of statistical evaluation of facts.1 must demonstrate how sufﬁcient reliability may be attained. 12.3. to be undertaken by staff of appropriate calibre specially trained for it. 12. Mortgages Reliable sequences of ﬁgures on the way prices have developed in the property market are essential for being able to calculate mortgages reliably. Development of such expert systems should therefore be left to software specialists in possession of estensive knowledge of numerical systems. the IT-supported method described here must be implemented by means of expert systems. too.2 and 8. This staff has to deal with the task areas. in that it has demonstrated that this is possible [ZK96].Final Considerations 143 calculating volatility rates on the basis of worst-case scenarios. and especially of the iterative calculations. with the result that today it regularly publishes a property market index for the Canton of Z¨ rich u (see Appendix 3).1 illustrated that there must be differentiation ˆ ˆ ˆ ˆ between four cases of the probabilities a.3 IT Support We have already mentioned. Here.4 Organisational Measures Putting this theory into practice calls for a comprehensive project. b. outlined above. These ﬁgures must be regularly ascertained for every borrower of a covered loan. and d occurring for the determination of the correlation. empirical tests as described in Section 1. in order to reduce their loan costs.

The costs of IT and of the project team would. is not least a question of expectations.1. whether the introduction is at ﬁrst made partly by imputation. So the risk is rather that the method may not deliver as much as it seems to promise. handling complex special cases. remain more or less unchanged. This.1. as described in Section 12. One way of reducing costs could lie in restricting implementation to just one part of the bank’s operations (for example to one type of loan transaction. could be limited. 12. however. the following preconditions must be fulﬁlled: r The theory must be corroborated empirically. then it must in actual fact be taking a de facto decision to bring the method in. That way costs. or wholly in transactions with customers.2 must be fulﬁlled. If a bank does decide to undertake empirical testing. as companies in the insurance industry already know all too well. This was indeed pointed out in Section 12. Many other examples that are not set down here have conﬁrmed this impression. We make no secret of the fact that banks will have to reckon with considerable costs. are likewise considerable. and ongoing development of the IT-based aids. applications for loans should still be examined in the conventional manner and then checked by the method presented here. The crux of the matter here is the following: the theory must already be introduced de facto in order for it to be possible to test it empirically. interest rates that are higher than the minimum rates are certainly permissible (cf. however. So it is permissible at this juncture to assume that absolute failure is ruled out. r The operational preconditions as described in Section 12. . particularly for specialist staff and if need be for the working up of statistical bases. or within one limited geographical area).6). from the point of view of expenditure. It remains therefore a matter of weighing costs against beneﬁts. This does not jeopardise the empirical test in any way at all: under the theory put forward here.3 APPLICABILITY OF THE METHOD PRESENTED To be able to apply the method presented. and on the basis that it should be the higher rate of loan interest that is applied to the commercial transaction with the customer. But in these cases the distinction must be drawn in operational accounting between risk premiums necessary according to the model and risk premiums that are in practice charged and collected. anyway. Loans should then only be granted if both methods of operation come to a positive result. however. The question then arises: how big is the danger of failure and what does it cost? The results of the theory are plausible in qualitative terms. a permanent specialist staff of mathematicians and statisticians must be set up. Section 4. As such a bank would have to be building up the necessary IT structures from scratch. The tasks for this staff include development of the theory. In order to limit the risk of mistakes arising from the theory. Implementation of this method would presumably be at its most simple for a bank entering into the business of lending for the ﬁrst time. it might do so as suggested here. On the other hand. as has already been mentioned at the beginning of Section 12.144 Risk-adjusted Lending Conditions Once the project work is concluded. keeping the statistical bases permanently up to date.1. Here it is not so important. the possible beneﬁts of really getting the costs of credit risks under control.

it will do this in its own interest. it is assumed here that between about ﬁve and 10 years must be allowed. . It is at this point that we venture to forecast that the banks that will be successful in the long term. why it has reached the conclusions it has reached. It will not necessarily do this if more favourable counter offers are available.5 OPEN QUESTIONS As we mentioned in Section 1. the time it takes for reliable facts to accumulate. in the same way as the concept of insurance is applied. It should not therefore be ruled out that a bank customer might not be prepared to make the efforts demanded of it. especially the ﬁnancing of projects. on the strength of an assessment of the position — in which the overall current and future business potential of this customer and the credit risks involved with it are weighed against each other — whether or not it is willing to persevere with its offer and its demands for information. as in the last analysis it jeopardises its own existence if it makes too many concessions in respect of its price and demands for information. in the business of lending. In particular we did not cover the following in terms of separate business transactions: r borrowers in the public sector and/or ratings related to countries r loans in foreign currencies r international lending business.8: [BCBS99. in the interest of a longer-term conﬁdence-building business relationship. The bank has to make the borrower aware.Final Considerations 145 This section concludes with a strong warning against making empirical testing — out of cost considerations — no more than a ‘trawl’ through the archives. will be those which can indeed say no — when their clients obtain more favourable counter offers from other banks — provided that they themselves are in a position to calculate risk-adjusted loan conditions correctly. 2]).10. The bank has to decide.4 CUSTOMER CONSIDERATIONS As already mentioned in Section 10. the borrower decides. From this it will follow that the borrower will have to reﬂect whether it always wants to go for the cheapest offer — with the disadvantage that this may be associated with frequent changes of bank partners — or whether it is on occasion prepared.2. 12. But it should also be taken into account that know-how will be building up continually — from ongoing evaluations — until the situation we are aiming for is reached. One must not underestimate. in such situations. to meet the bank’s information requirements and to pay the price the bank has asked. either. If it is sure that it has calculated the loan interest rate correctly. something is both demanded and expected of the borrower when this model is applied. Our own experiences with such exercises show that (a) they do not as a rule come up with the desired results. S. the subject of this study was deliberately limited. and (b) they do not really save any money. According to the terms of the loan products to be examined. The Basel Committee itself reckoned on several years (cf. Section 1. As we explained in Section 1. whether or not it is prepared to pay the risk-adjusted price being asked.3. 12.

and the ﬁeld for further studies is therefore still wide open. [MANZ98] and the reading suggested there). Many details still await clariﬁcation. it has only been possible to give an overall view. Efforts are being made to develop methods that scale down the shortfall risk of a whole portfolio of loans vis-` -vis the a sum of the individual loan risks (cf. for example. To what extent the method we have introduced here may contribute anything in this ﬁeld must be the subject of future studies. So it is as true as ever that providing the answers to old questions all too often forms the starting point for posing new ones! . In portraying the method we have introduced.146 Risk-adjusted Lending Conditions Furthermore the subject of portfolios of loans was likewise not covered. but it has been necessary to conﬁne ourselves to a small selection. The meaningful content of this theory has thus not yet been fully developed by any means. 12.6 CLOSING REMARKS Some idea of the consequences of the method we have introduced here may have been communicated with the help of illustrations and examples.

43 7.41 7.1 4.3 7.1 7.56 7.1 2.4 ˆ a b ˆ b bc ˆ c d ˆ d e f g i i(t) i c (t) id ig im i mt i pa is i spa p r rc rcac ˆ r t v A B probability breakdown distribution rate probability probability corrected breakdown distribution rate probability probability debt rate probability return on equity ﬁnancing cost rate return rate loan interest rate loan interest rate according to the whole term corrected loan interest rate according to the whole term discount rate return on a risk-free investment in government bonds marginal interest rate return on the market interest rate on annual basis risk-free standard interest rate annual standard interest rate proﬁt contribution rate shortfall risk hedging rate correction factor current account credit write-off risk hedging rate correlation coefﬁcient term return on assets number of trading days per year breakdown distribution probability value .2 7.50 7.21 8.37 9.21 1.40 1.21 4.4 8.1 7.43 9.30 8.56 8.2 7.22 1.1 9.2 2.Appendix 1: Notation Equation number 8.20 8.2 9.40 1.1 1.2 7.

10 7.1 7.1 2.1 7.28 1.2 7.43 2.4 4.148 Risk-adjusted Lending Conditions Bc C Cj D E L Lg Lu N P Rb Rn S V VI X β χ χ∗ ε ϕ x µ ρ ρ∗ ρa ρc ρB ∩ C σ ψj Γ Λ corrected breakdown distribution free cash ﬂow cash ﬂow face value after j periods debts equity payed out loan amount granted loan used loan standard normal distribution function put value billed return necessary return proportional salaries value of the company liquidation value portfolio values measurement of unlevered assets market risk according to CAPM survival chance survival chance regarding the breakdown distribution probability value rate cumulative success chance cumulative shortfall risk credit-worthiness key ﬁgure medium of the logarithms shortfall risk credit shortfall risk average of all n shortfall risks ρk corrected credit risk combined risk volatility of the company value according to the time period probability of cash ﬂow Cj gamma function loan market value 7.20 4.7 5.2 2.2 4.1 .19 7.45 4.8 3.28 7.8 7.57 8.2 2.45 2.1 1.2 3.19 4.41 1.2 7.54 7.46 8.2 7.54 7.20 7.2 7.

] A 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Debts L tot Amount of loan L Term t in years Standard interest rate i s Market value of the company V Volatility of the market value σ Privileged salary/wage claims S Mortgage d Volatility according to term t x N (x − σ (t)) N (x): probability of bankruptcy ρ Loan risk ρ ∗ Loan risk per annum ρ pa (Input ﬁeld) (Input ﬁeld) (Input ﬁeld) (Input ﬁeld) (Input ﬁeld) (Input ﬁeld) (Input ﬁeld) = B1/B5 = B6∗ SQRT(B3) B = LN(B8/(1-B13))/B9+B9/2 = STANDNORMDIST(B10-B9) = STANDNORMDIST(B10) = (B11-B8∗ B12)/(B11-B8) 1-(1-B13)ˆ(1/B3) (Continued) . [For this iterations absolutely must be allowed under Extras/ Options/Calculations.Appendix 2: Excel Worksheet An Excel Worksheet set up as follows proved to be the best way of calculating the equations in Chapter 7.

1. d ≤ 0 or d ≥ 1.e. Usually 10 will sufﬁce. Hint 3: In the case of unreliable ﬁgures for d (Field B7): i. If this happens click on Field B13.1 may indeed be programmed in Excel.150 Risk-adjusted Lending Conditions 15 16 17 18 19 20 21 22 23 24 25 Loan interest rate i pa Breakdown distribution rate b Loan demand at maturity Breakdown distribution probability value B Proportional privileged salary/wages claims S Corrected value of B: Bc Corrected breakdown distribution rate probability bc ∗ Corrected loan risk ρc ∗ Corrected loan risk per annum ρcpa (B4+B11)/(1-B14) 1-B13/B12 B2∗ (1+B15)ˆB3 B16∗ B17 B7∗ B2/B1 IF(B18>B19. Table 2. Reducing the maximum number of iterations under the presetting is therefore recommended. Hint 2: On occasion Excel manifests difﬁculties with this worksheet and reacts with various fault reports in the B column. and then press ENTER. but this is very laborious. and then F2. Hint 4: If only the loan amount in line 2 is changed from one calculation to the next.0) B20∗ (1-B12)/(B2∗ (1+B4)ˆB3-B20∗ B12) B12∗ (1-B21) 1-(1-B22ˆ(1/B3) (B4+B23)/(1-B23) B24-B15 Corrected loan interest rate i cpa Corrected loan interest rate on account of S Hint 1: The loan interest rates in B15 and B24 are calculated on the basis of the loan risks calculated in B14 and B23.B18-B19. . then only the values in lines 21 to 25 change in the results. can be seen under EXTRAS/OPTIONS/CALCULATIONS) and then aborts with fault reports. and not on the basis of the shortfall risk of the rating level concerned according to Table 2. Excel runs through the maximum number of iterations (presetting usually 100.

6 189.9 161. Year 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 Single-family home 100.6 168.5 110.9 110.3 287.8 226.6 119.2 192.1 144.9 178.6 221.0 190.4 120.8 167.0 115.6 192.6 Condominium 100.9 216.0 143.2 188.0 103.8 212. dated 5 May u u 2000.zkb.2 243.9 171.6 225.0 124.7 219.0 200.7 167.0 179.9 218.5 245.2 Multiple dwelling unit 100.7 196.3 147.9 188.0 138.3 171.5 150.5 223.6 208.1 148.1 183.1 132.7 160.4 161.6 129.2 222.0 214.1 113.3 200.6 193.0 102.4 156.5 124.9 193.8 162.Appendix 3: Property Price Index The Cantonal Bank of Z¨ rich’s property price index for the Canton of Z¨ rich.7 http://www.ch/bin/entry/frame/private/immobilien/index.html Source: Cantonal Bank of Z¨ rich u .5 211.

.

in each case. For a borrower deﬁnitely either to default or not to default in the second year. 324] the following applies: ϕ2 = χ1 · ρ2 ε2 = χ1 · χ2 The same applies.5) (3. for further considerations. j=1 = j−1 k=1 (3. for the following periods: ϕ j.7) (3.8) may be simpliﬁed as follows: ϕ j = χ ( j−1) · ρ = (1 − ρ)( j−1) · ρ if ρ1 = · · · = ρn = ρ εj = χ j if χ j = · · · = χn = χ (3.8) χk On the assumption that ρ j and χ j have the same value.7) and (3.3) (3. S. the following self-excluding events generally [BOHL92. According to the laws of multiplication of probability calculus [BOHL92. it may not default in the ﬁrst year. As the default may only occur in one and not in several periods. S. that ρ j+1 and χ j+1 are independent of ρ j and χ j .Appendix 4: Chapter 3 — Derivations First of all we calculate the values of ϕ j and ε j for each individual period of the n periods.10) Under the deﬁnition above ϕ(n) means the probability that the borrower will default not in any speciﬁed but in any one of the n periods. The solution for the ﬁrst period is evident: ϕ1 = ρ1 ε1 = χ1 (3.9) (3. if it occurs at all. .4) It is assumed.6) χk · ρj = εj = j k=1 j−1 k=1 (1 − ρk ) · ρ j (3. for all n periods. by analogy. equations (3.

as in contrast to shortfall.18) and. success only occurs if each individual period of the n periods was successful.15) The following thus applies in the special case: ϕ(n) + ε(n) = (1 − χ n ) + χ n = 1 which equation (3.11) and.17) Reﬂection now leads to the probability ψ j . especially: ψ j = χ j = (1 − ρ) j if χ1 = · · · = χn = χ (3.13) The law of addition does not apply to ε(n). especially: n ϕ(n) = j=1 (1 − ρ)( j−1) · ρ if ρ1 = · · · = ρn = ρ (3. under which the following applies: ϕ(n) = ρ · 1 − (1 − ρ)n = 1 − (1 − ρ)n = 1 − χ n 1 − (1 − ρ) (3. ε(n) thus corresponds to εn (cf. that ψ j is none other than the probability that no loss arises within the ﬁrst j periods.12) According to equation (3. So in general the following applies: j j ψj = εj = k=1 χk = k=1 (1 − ρk ) (3.19) .154 Risk-adjusted Lending Conditions 313] apply according to the laws of addition of probability calculus: j n j−1 ϕ(n) = j=1 ϕ j = ρ1 + j=2 k=1 (1 − ρk ) · ρ j (3. especially: ε(n) = χ n = (1 − ρ)n if χ1 = · · · = χn = χ (3.14) and.10)). So in general the following applies: n ε(n) = j=1 χj (3.16) ϕ(n) = 1 − j=1 χj = 1 − (1 − ρ j ) j=1 (3. equations (3. In the general case the following results: n n (3.12) ϕ(n) is a geometric series with a1 = ρ as its ﬁrst term and the multiplication factor q = (1 − ρ).2) conﬁrms.8) and (3.

Λ = L applies. equation (4. Reducing with L and putting i outside the brackets in the ﬁrst summand results in: n 1=i· j=1 χj (1 + i s ) j + χn + (1 + i s )n n j=1 χ j−1 · ρ · b · 1 + i 1 + is j (4.2) is rewritten as follows: 1= ρ · b · (1 + i) · (1 − ρ n ) i · ρ · (1 − ρ n ) + ρn + (1 − ρ) (1 + i s ) · (1 − ρ) ρ · b · (1 + i) · (1 − ρ n ) (1 + i s ) (4. This now results in: ρ · b · (1 + i) (4.1 DERIVATION According to the assumptions that were made (cf. we are likewise concerned with a geometric series with: a1 = ρ= (4. By using the overall equation for geometric series and substitution by equation (4.7). the same risk-free rate of interest for discounting is used for each summand.4) So ρ is identical in both geometric series and only a1 is different. Section 1.5) Taking the mean summand ρ n on the left-hand side and extending by (1 − ρ) results in: (1 − ρ) · (1 − ρ n ) = i · ρ · (1 − ρ n ) + (4. we are concerned with a geometric series with: χ a1 = ρ = (1 + i s ) ρ · b · (1 + i) (1 + i s ) χ (1 + i s ) (4.7) (1 − ρ) = i · ρ + (1 + i s ) .2) In the case of the ﬁrst summand. is important.3) In the case of the second summand. Furthermore.Appendix 5: Chapter 4 — Derivations SECTION 4.3).6) The fact that the equation may be shortened at this point by (1 − ρ n ).

11) As r is independent of n.15) As χ ≤ 1 always applies.8) Reverse substitution of ρ and replacement of i by (i s + r ) according to equations (1.3.3) results in: 1− χ = (i s + r ) · (1 + i s ) χ +ρ·b 1 + is + ρ·b (1 + i s ) (4.9) Multiplication by (i s + r ) and replacing χ by (1 − ρ) results in: i s + ρ = (i s + r ) · (1 − ρ + ρ · b) + ρ · b (4. in which the expectation value of the repayment of capital according to Section 3. The key result for the shortfall risk hedging rate r comes thus to be independent of the term of the loan! Tasking i out of brackets results in: (1 − ρ) = i · ρ + ρ·b (1 + i s ) + ρ·b (1 + i s ) (4.16) (4.10) Multiplying out and insertion of (1 − ρ + ρ · b) = (1 − ρ ∗ ) (see equation (2. as there is in practice no such thing as ‘free credit’.13) (4.15) turns into: 1=i· ρ · b(1 + i) ρ + 1−ρ (1 + i) · (1 − ρ) ρ · b · (1 + i) (1 − ρ) = i · ρ + (1 + i s ) (4.156 Risk-adjusted Lending Conditions So the path of the term with the time element of the number of periods n has been shortened as a consequence of the assumption that shortfall risks are constant.6)) results in: i s + ρ = i s · (1 − ρ ∗ ) + r · (1 − ρ ∗ ) + ρ · b Solution by r results in: r= i s + ρ − i s · (1 − ρ ∗ ) − (ρ · b) (1 − ρ ∗ ) ∗ is + ρ − is + is · ρ ∗ r= (1 − ρ ∗ ) ∗ ρ r= · (1 + i s ) 1 − ρ∗ (4. thus: 1=i· ρ · b(1 + i) ρ + ρ∞ + 1−ρ (1 + i s ) · (1 − ρ) (4. ρ < 1 always applies on the assumption that i > 0 always applies s likewise. This means that ρ ∞ is a null consequence and equation (4. This assumption is permissible.1 is precisely zero! This can be veriﬁed.5) is written for a limitless geometric series with a limitless number of periods. the same shortfall risk hedging rate r must also be valid for a limitless number of periods of loan term.14) (4.17) . in that equation (4.1) and (1.12) (4.

35) . The value ρ ∗ = ρ(1 − b) (4.33) Reformulation of the ﬁrst and third summands gives: λ= i · χl · 1 − χll (1 + i sl )l (1 + i sl − χl ) ρl · b · (1 + i) · 1 − (1 + lisl )l χll + + (1 + i sl )l (1 + i sl − χl ) χll (1 + i sl )l −1 (4.30) there results: l j j−1 (4. there results: λ= i· χl (1 + i sl ) · 1− χll (1 + i sl )l 1− χl (1 + i sl ) ρl · b · (1 + i) · 1 − (1 + lisl )l χll + + χ (1 + i sl )l (1 + i sl ) · 1 − (1 +lisl ) χl χl −1 (4.17) are identical and thus must lead to the same conclusion.30) L =Λ−L = j=1 χl · i · L (1 + i sl ) j + χll · L + (1 + i sl )l l j=1 χl · ρ · b · L · (1 + i) (1 + i sl ) j −L (4. SECTION 4.7) and (4.5 DERIVATION The deviation L from the nominal amount is calculated as the difference of: L =Λ−L This means: L > 0 appreciation proﬁt L < 0 need for provision to be made From (4.18) is none other than the credit shortfall risk. which is dependent on the shortfall risk ρ of the borrower and on the probable breakdown distribution rate b.34) Giving them a common denominator results in: λ= i · χl · (1 + i sl )l · 1 − + χll · (1 + i sl − χl ) χll (1 + i sl )l (1 + i sl − χl ) · (1 + i sl )l + ρl · b · (1 + i) · (1 + i sl )l · 1 − − (1 + i sl − χl ) · (1 + i sl )l (1 + i sl − χl ) · (1 + i sl )l (4. we are again concerned with geometric series.31) Abbreviation using L results in: L =λ= L l j=1 χl · i (1 + i sl ) j j + χll + (1 + i sl )l l j=1 χl j−1 · ρl · b · (1 + i) (1 + i sl ) j −1 (4.29) and (4. By using the appropriate overall equations.32) In the case of both summands.Appendix 5: Chapter 4 — Derivations 157 The comparison shows that equations (4.

37) The two right-hand brackets in the summands of the numerator are identical up to the reversed plus/minus sign. other terms in these brackets may be put together. so that the following conclusion emerges: λ= l 1 + i sl − χll · [(1 + i) · (ρl · b + χl ) − (i sl + 1)] (1 + i sl )l · (1 + i sl − χl ) (4.158 Risk-adjusted Lending Conditions Multiplying them out in the numerator results in: λ= l 1 + i sl − χll + χll · (1 + i sl − χl ) (1 + i sl − χl ) · (1 + i sl )l l ρl · b · (1 + i) · 1 + i sl − χll − (1 + i sl − χl ) · (1 + i sl )l + (1 + i sl − χl ) · (1 + i sl )l i · χl · (4.36) Taking (1 + i sl )l and χll out of brackets gives: λ= (1 + i sl )l · (i · χl + ρl · b · (1 + i) − 1 − i sl + χl ) (1 + i sl − χl ) · (1 + i sl )l l χ · (1 + i sl − χl − i · χl − ρl · b · (1 + i)) + l (1 + i sl − χl ) · (1 + i sl )l (4. − is the valuation correction in percentage terms in relation to the nominal value of the loan. and may thus be taken out of brackets. In addition. .38) According to equation (4.32).

50]: (1 + x)a ≈ 1 + ax can be written: 1 − nρa − y≈ (1 + n (5. as what we are concerned with here is the total of the deviations from the mean: n ρk = 0 k=1 (5. By applying the approximation formula [DMK/DPK92.15) leads to: y≈ 1 − ρa 1 + is n (5.9) can be written as .10) should be simpliﬁed for y. equation (5. by way of approximation.12) in the numerator of formula (5.12) ρk (5.15) Renewed application of the approximation equation (5. permissible to use this approximate solution also for only j periods instead of for all n periods.16) On the assumption that it is.Appendix 6: Chapter 5 — Derivations SECTION 5. S.13) k=1 i s )n The summand in the numerator is by deﬁnition equal to zero.14) So one obtains: y≈ 1 + nρa (1 + i s )n (5.2 DERIVATION First of all equation (5.

25) .17) and (5.160 Risk-adjusted Lending Conditions follows: n x≈ j=1 1 − ρa 1 + is j (5.22) can be written as follows: z≈ ρ · (1 − ρ n ) ρa · 1 − ρa 1−ρ (5. The sum in equation (5. by way of approximation.20) 1 − ρa 1 + is (5. equation (5. (5.17) is a geometric sequence with: a1 = ρ = This leads to the following result: x≈ ρ · (1 − ρ n ) 1−ρ (5.11) can.20). be written as follows: n z≈ j=1 (ρa + ρ j ) · (1 − ρa ) j−1 = (1 + i s ) j n j=1 (ρa + ρ j ) · (1 − ρa ) 1 − ρa 1 + is j (5.22) On the assumption that ρ j may.19) equation (5.21) By analogy with equation (5.21) and (5. be ignored — also by way of approximation — and by using equations (5.17).24) Reduction by (1 − ρ n ) and extension by (1 − ρ) results in: i≈ 1−ρ− ρ+ b · ρa · ρ (1−ρa ) b · ρa · ρ (1−ρa ) = 1−ρ· 1+ ρ· 1+ b · ρa 1−ρa b · ρa 1−ρa (5.16) can be written as follows: y ≈ ρn (5.6) results in: i≈ 1 − ρn − b · ρa · ρ · (1−ρ n ) (1−ρa ) · (1−ρ) ρ · (1−ρ n ) ρ · (1−ρ n ) + (1−ρa ) · (1−ρ) (1−ρ) (5.23) into equation (5. when adding up.19) equation (5.17) Here it is presupposed that for all values of j j ρk ≈ 0 k=1 (5.23) The insertion of equations (5.18) applies with sufﬁcient precision.19) By use of the substitution (5.

49) may thus — by way of approximation — likewise be used.32) {” ”} = {” ”} = {” ”} = i · n i · n i · 1+ b·ρ j 1−ρ j + b · ρj (1 − b j ) n · (n − j + 1) + n b · ρj b · ρj + 1 − ρj (1 + ρ j ) · i b · ρj 1 − ρj · 1+ 1 i · (n − j + 1) + · (n − j + 1) + 1 i 1 i b · ρ j · (n− j+1) 1−ρ j (5.30) may now. only the curved brackets are reformulated at ﬁrst: {” ”} = i + b · ρj b · ρj +i · 1 − ρj 1 − ρj b · ρj 1 − ρj · n− j +1 1 + n n · [n − j + 1] + 1 (5.29). in that the average ρa replaces the constant ρ.27) Extension by (1 + i s ) and taking ρa out of brackets gives: i≈ i s + ρa · (1 − b) 1 + i s − 1 + ρa · (1 − b) = 1 − ρa · (1 − b) 1 − ρa · (1 − b) ∗ i s + ρa i≈ ∗ 1 − ρa (5. be summarised as follows: i · L+ b · ρ j · (1 + i) (1−ρ j ) n · L · 1− L= j=1 ( j − 1) n + L n · j k=1 (1 − ρk ) (5. SECTION 5.4 DERIVATION Equation (5.36) i · 1+ 1+ {” ”} = 1+ . In the case of the variable ρ j . equation (4.Appendix 6: Chapter 5 — Derivations 161 Reverse substitution gives: i≈ Multiplying out gives: i≈ 1− 1 − ρa + b · ρa (1 + i s ) 1 − ρa + b · ρa (1 + i s ) 1− 1 − ρa 1 + is 1 − ρa 1 + is · 1+ · 1+ b · ρa 1 − ρa b · ρa 1 − ρa (5.28) (5.31) (1 + i s ) j After abbreviation by L.49) reveals that this is identical to equation (5.33) +1 (5.35) (5.34) (5.26) (5. owing to the symmetries.29) Comparison with equation (4.

30) abbreviated by L.38) (1 − ρk ) .162 Risk-adjusted Lending Conditions Inserted into equation (5. and reformulated results in: n = i n j=1 1+ b · ρj 1 − ρj · 1+ 1 i · (n − j + 1) + (1 + i s ) j n 1 i · j k=1 (1 − ρk ) (5.37) and transposed into: i= n j=1 1+ b·ρ j 1−ρ j · 1+ 1 i · (n − j + 1) + (1 + i s ) j 1 i · j k=1 (5.

: Principles of Corporate Finance. Fabozzi. Robert: Problem Loans and Cost Efﬁciency in Commercial Banks. 47 (1). Bakshi. Frankfurt am Main 1991. E. S. Bern.htm). Wien. Altmann. Varenne de. Brealey. Lea V. E. Haldeman. Investment.1996. Stuttgart. Fran¸ ois: Valuing Risky Fixed Rate Debt: An Extension.Auﬂage. in: Journal of Banking & Finance. in: Journal of Finance. Fischer. Peter: Statistik. New York 1996. E. G. 44. Zhiwu: Empirical Performance of Alternative Option Pricing Models. Everling.und Sozialwissenschaftler. and Corporate Value: an Empirical Analysis. Christoph: Bank¨ bernahmen in der Schweiz. in: Journal of Lending & Credit Risk Managment. Richard A. in: Journal of Political Economy. 19. 1997.. DeYoung. February 1998. June 1997.: Black–Scholes and Beyond — Option Pricing Models.. Frank J. Narayana. Cho. Roland: Der Firmenkunde will mehr. Berger. Stuttgart.bis. u Briys. 29–54. 21. Wien 1994. u Brakensiek. in: Journal of Financial Economics. Basle Committee on Banking Supervision: Credit Risk Modelling: Current Practices and Applications. u u 5. B¨ schgen. Hans E. Boeschenstein. Wiesbaden 1996. Thomas: Die Kalkulation und Steuerung von Ausfallrisiken im Kreditgesch¨ ft a der Banken. Narayana. Basel 1999 (http://www.Bibliography ALT89 ALTM77 AUCK94 BCBS99 BCCH97 BEDE97 BLSC73 BOES96 BOHL92 BRAK91 BRMY96 BRUN94 BRVA97 ¨ BUEV CAFA92 CAOU98 CART98 CHO98 CHRI97 Altman. Allen N. B. I. Chriss. Caouette. Stewart C. in: Handelszeitung Nr. Wien 1994. Charles. M¨ nchen. R.. Oldenburg 1992. Cen.: Managing Credit Risk: the Next Great Financial Challenge.: Zeta Analysis: a New Model to Identify Bankrupty Risk of Corporations. I..09. J. Myeong-Hyeon: Ownership Structure. Myers. Black. Einf¨ hrendes Lehrbuch f¨ r Wirtschafts. Chicago/ London/Singapore 1997. P. 52 (2). Oliver (Herausgeber): Handbuch Rating. u Carlson..org/wnew. P.: The Trading and Securitization of Senior Bank Loans. December 1997. in: Journal of Banking and Finance. Gurdip. Brunner.. ﬁfth edition. I. Chicago. January 1998. Carty. 103–121. John H. Z¨ rich 1996. Cambridge 1992. u Bohley. 1989.. Scholes. Neil A. Cao.: Measuring Corporate Bond Mortality and Performance. . Myron: The Pricing of Options and Corporate Liabilities. New York 1998.: Credit Risk and Credit Quality Correlation: What Can We Measure?. in: Journal c of Financial and Quantitative Analysis.. in: Journal of Finance. Christoph: Finanzmathematische Grundlagen des Investment Banking. 38. Eric. Bern. Auckenthaler. 32 (2). May–June 1973. Altmann. June 1977.

Pirotte. Singer. DMK/DPK92 Deutschschweizerische Mathematikkommission. Robert. Futures. forthcoming. Publikation der DG Bank. Cambridge 1992. E. John C. Lune.: The Value of Corporate Debt with a Sinking-Fund Provision. Schreiber. 19. in: Risk.: Options. Krob Bernhard: Kundenindividuelle Bewertung des Ausfallrisikos mit dem Optionspreismodell.154). Lloyd.cob. in: Journal of Finance.: On Measuring Credit Risks of Derivative Instruments. GREN96 Grenadier. DUFF95a Duffee.ohio-state. in: Strategisches Bankcontrolling — Eine neue Herausforderung. GLAN94 Glantz. FHSS94 Flesaker. February 1995. HOF95 Hof. 42 (3).cob. Frank R. D. Ramsey. Rubinstein. Cambridge 1994.. Frankfurt am Main. http://www. John: The Seven Stages of Risk Management.1995 im Rahmen der IIR-Fachkonferenz “Kreditmanagement”. Werner: Risiko — aus morphologischer Sicht. R. 46. in: Journal of Finance.. 5. Wien 1994. Hughston.. Upper Saddle River 1997. J. Carey. November 1996. Federal Reserve Board.: The Price of Default.164 COME91 COPI01 CORU85 CPSH98 Risk-adjusted Lending Conditions Cooper. ¨ HURL93 H¨ rlimann. Thomas S. Ronald F. Federal Reserve Board. in: Working Paper. Auﬂage. Taking all the Credit. Laurence.edn/htbin/htimage/˜ﬁn/journal/jf. in: Journal of Finance.conf?94. A.: The Valuation of Corporate Liabilities as Compound Options: A Correction. Marz. u Z¨ rich 1993. Mello.conf?94. Sprung.. Frank J. Ramirez.cob. Sartain. Steven R. in: Journal of Financial and Quantitative Analysis.. FRSK97 Fooladi.. Bjorn. Gary L. Englewood Cliffs 1985.conf? 94. Z¨ rich 1994. June 1998 (http://www. in: Working Paper.. R. 1991. u HENN94 Henninger. Mark. GEKR Gerdsmeier. IIia D. third edition. in: Journal of Business. FRRM97 Fabozzi. in: io Management Zeitschrift. G. DICH98 Dicher. White A. DRZI98 Drzik. Mitch. H. 19. Mark: Options Markets. Roberts. in: Journal of Finance. Chicago. Philipp: Neuere theoretische Grundlagen zur Fundierung eines modernen u Kreditmanagements. G. February 1998. Chuck. in: Journal of Lending and Credit Risk Managment. HULL97 Hull. Stefan.und Arbeitstagung des Instituts f¨ r schweizerisches a u Bankwesen der Universit¨ t Z¨ rich. in: Risk. Y. 1984. . Bern. Part. 53. Cossin. in: Journal of Financial and Quantitative Analysis. HOSI84 Ho.ohiostate-edu/htbin/htimage/˜ﬁn/ journal/jf.: Dictionary of Financial Risk Management. u HUWH92 Hull. Chicago. Stuttgart. 21.: Determinants of the Consumer Bankruptcy Decision.: The Default Risk of Swaps. la. September. 1997. Stevan A: Does Corporate Lending by Banks and Finance Companies Differ? Evidence on Specialization in Private Debt Contracting. H. I. HALB94 Halbherr. GAST92 Gastineau. Deutschschweizerische Physikkommission: Formeln und Tafeln Mathematik — Physik.: Leasing and Credit Risk. Sharp.ohio-state. New Hope PA 1997. Z¨ rich 1992.. Robert L. Iraj J. DUFF95b Duffee. Publikation der DG Bank. Robert: The Valuation of Corporate Liabilities as Compound Options. John C. 1992. 1984.154. Morton: Loan Risk Management. u DOSA97 Domowitz. and Other Derivatives. Ute: Branchenrisikoeinsch¨ tzung im Rahmen der Bonit¨ tsbeurteilung von a a kommerziellen Kreditkunden.. GESK74 Geske. in: Journal of Financial Economics.09.edn/htbin/htimage/˜ﬁn/ journal/jf. Shinner Frank: Duration for Bonds with Default Risk. 1984. Vortrag am 14. Frankfurt am Main. Gordon S. June 1998 (http://www. in: Informations.154). Chichester 2001. 57. 7 (9) September 1994. GEJO84 Geske. 53.: On Variation of Default Risk with Treasury Yields. Johnson.: Advanced Credit Risk Analysis. in: Journal of Banking & Finance. Michael: Kreditderivate — Aktive Steuerung von Kreditrisiken. Cox.: Is the Risk of Bankruptcy a Systematic Risk?. Michael: The Handbook of Nonagency Mortgage-Backed Securities. January 1995. 62.

Bibliography

165

HUWH95 Hull, J.; White A.: The Impact of Default Risk on the Prices of Options and other Derivative Securities; in: Journal of Banking and Finance, 19 (May); 1995. HSIA81 Hsia, Chi-Cheng: Optiomal Debt of a Firm: An Option Pricing Approach; in: Journal of Financial Research, 4; 1981. JATU95 Jarrow, R. A.: Turnbull, S. M.: Pricing Options on Derivative Securities Subject to Credit Risk; in: Journal of Finance, 50; 1995. JOST87 Johnson, H.; Stulz, R.: The Pricing of Options under Default Risk; in: Journal of Finance, 42; 1987. KAEL98 Kaelhofer, Stephen: Portfolio Management of Default Risk; in: http://www.kmv.com; 1998. KARO93 Karolyi, Andrew G.: A Bayesian Approach to Modeling Stock Retrun Volatility for Option Valuation; in: Journal of Financial and Quantitative Analysis, 28 (4), December 1993. KILG94 Kilgus, Ernst. Strategisches Bank-Management, Bern, Stuttgart, Wien 1994. ¨ KNAB95 Knabenhans, Walter: Die Transparenz im Markt muss erh¨ ht werden — Uber die Zukunft, o Chancen und Risiken des Derivativmarktes; in: Finanz und Wirtschaft Nr. 18, 04.03.1005; Z¨ rich 1995. u KREY91 Kreyszig, Erwin: Statistische Methoden und ihre Anwendung, 7. Auﬂage; G¨ ttingen 1991. o KRRO92 Kremer, Joseph W.; Roenfeldt, Rodney L.: Warrant Pricing: Jump-Diffusion vs. Black– Scholes; in: Journal of Financial and Quantitative Analysis, 28 (2), June 1992. LEE81 Lee, C. Jerons: The Pricing of Corporate Debt: A Note; in: Journal of Finance, 36, 1981. MANZ98 Manz, Felix: Prozessorientiertes Kreditmanagement; Bern, Stuttgart, Wien 1998. MASU81 Mason, Scott P.; Sudipto, Bhattacharya: Risky Debt, Jump Processes, and Safety Covenants; in: Journal of Financial Economics, 9, 1981. MEIE96 Meier, Christian: Lehren aus Verlusten im Kreditgesch¨ ft Schweiz; bern, Stuttgart, Wien 1996. a MERT73 Merton, R. C.: Theory of Rational Option Pricing; in: Bell Journal of Economics and Management Science, 4, Fr¨ hling 1973. u MERT74 Merton, R. C.: On the Pricing of Corporate Debt: the Risk Structure of Interest Rates; in: Journal of Finance, 2, 1974. MOOD90 Moody’s Investors Service: Einf¨ hrung Credit-Ratings am Kapitalmarkt; Frankfurt, London, u New York, Paris, San Francisco, Sydney, Tokyo 1990. ¨ MULL90 M¨ ller, J¨ rg: Branchenkennzahlen zur Bonit¨ tsbeurteilng von Klein- und Mittelunu o a ternehmungen — Eine empirische Studie anhand von Zahlenmaterial der SKA; bankinterne Studie der SKA; Z¨ rich 1990. u NATE94 Natenberg, Sheldon: Option Volatility and Pricing; Chicago, Cambridge 1994. NEWT93 Newton, Brian: Modeling Credit Risk; New York 1993. NZZ96 – Acra statt Abracadabra, Das neue Risikosystem des Schweizerischen Bankvereins; Neue Z¨ rcher Zeitung, Nr. 218, Seite 29; Z¨ rich 1996. u u – Altlasten-Beseitigung bei der CS Holding; Neue Z¨ rcher Zeitung, Nr. 298, page 19; Z¨ rich u u 1996. PASS94 Passardi, Adriano: Bank-Management und Bank-Kostenrechnung; Bern, Stuttgart, Wien 1994. ¨ RASS93 R¨ ss, Hugo E.: Die Resrukturierung von Unternehmen aus der Sicht der Kreditgebenden a Bank; Bern, Stuttgart, Wien 1993. RITR98 Ritchken, Peter; Trevor, Rob: Pricing Options Under Generalized GARCH and Stochastic Volatility Process; in: Journal of Finance, forthcoming, http://www.cob.ohiostate.edu/htbin/htimage/˜ﬁn/journal/jf.conf?94,154. RSSO93 Reswick, Bruce G.; Sheikh, Aamir M.; Song, Yo-Shin: Time Varying Volatilities and Calculation of the Weighted Implied Standard Deviation; in: Journal of Financial and Quantitative Analysis, 28 (3); September 1993. SAVE96 Savelberg, Albert H.: Risikomanagement mit Kreditderivaten; in: Die Bank, 6, 1996. ¨ SCMU98 Scholl, Martin; M¨ ller, Roger: Risikogerechte Preise; in: Schweizer Bank; 1; 1998. u SCHW82 Schwartz, Eduardo S: The Pricing of Commodity-Linked Bonds; in: Journal of Finance, 37; 1982.

166 SDHJ97

Risk-adjusted Lending Conditions

Scott Docking, Diane; Hirschey, Mark; Jones Elaine: Information and Contragiven Effects of Bank Loan–Loss Reserve Announcements; in: Journal of Financial Economics, 43; 1997. SMIT76 Smith, Clifford W. Jr.: Option Pricing; in: Journal of Financial Economics, 3; 1976. SWDI92 Swidler, Steve; Diltz, David J.: Implied Volatilities and Transaction Costs; in: Journal of Financial and Quantitative Analysis, 27 (3); September 1992. TRTU96 Tripi, R.; Turban, E.,: Neural Networks in Finance and Investing; in: http://www.sigmaresearch.com/bookshelf/rtbookn2.thm. VASI84 Vasicek; Oldrich A.: Credit Valuation; in: http://www.kmv.com; 1984. VIZA96 Vintschger von, R.; Zaker, S.: Kreditrisiken sind gef¨ hrlicher als Marktrisiken; in: Finanz a und Wirtschaft, 21.09.1996; Z¨ rich 1996. u VOLK93 Volkart, Rudolf: Beitr¨ ge zur Theorie und Praxis des Finanzmanagements, 5. erweiterte a Auﬂage; Z¨ rich 1993. u VOLK98/1 Volkart, Rudolf: Optionswertbildung und Optionspreistheorie — Versuch einer integrierten ‘didaktischen’ Darstellung; in: Working Paper Serie (Institut f¨ r schweizerisches Bankwesen u der Universit¨ t Z¨ rich), Z¨ rich 1998. a u u ¨ VOLK98/2 Volkart, Rudolf: Optionstheoretische Uberlegungen zur Kapitalstrukturgestaltung und zur Kreditﬁnanzierung; in: Working Paper Serie (Institut f¨ r schweizerisches Bankwesen der u Universit¨ t Z¨ rich), Z¨ rich 1998. a u u VOLK99 Volkart, Rudolf: Optionstheoretische Analyse von Fremdkapitalkontrakten und alternative Credit Risk Spread-Bestimmungen — eine integriete Darstellung; in: Working Paper Serie (Institut f¨ r schweizerisches Bankwesen der Universit¨ t Z¨ rich), Z¨ rich 1999. u a u u WONG97 Wong, Kit Pang: On the Determinants of Bank Interest Margins under Credit and Interest Rate Risks; in: Journal of Banking and Finance, 21; 1997. ¨ ZELL83 Zellweger, Bruno: Uberwachung kommerzieller Bankkredite; Bern, Stuttgart 1983. ZIMM98 Zimmermann, Heinz: C = S × N(d1 ) − X · e−r.(T−t) × N(d2 ); in: Schweizer Bank; 1998. ZKB96 Z¨ rcher Kantonalbank: Immobilienmarkt Z¨ rich; Eigenverlag; Z¨ rich 1996. u u u

Index

acceptability of debt servicing, 112–115 annual rate of interest, 66 annual repayment, regular, loans on, 36 annual volatility, 66–67 approximate solution, 48–49 reliability of, 49–50 assets, 54 return on, 108–110 volatility of, 121 assumptions, model, 11–12 balance sheets, analysis, 123 bankruptcy expectation value of loss in, 20–21 privileged salary claims and, 80–82 risk of, 78–79 bankruptcy-resistant debt rate, 115 Basel Committee for bank supervision, 12, 13, 145 behavioural risks, 3, 9 Black/Scholes approach cf KMV approach, 59–60 bonds, 80 breakdown distribution probability value, 80 breakdown distribution rate probability, 78 Bretton Wood agreements, 15 call put parity, 61 Cantonal Bank of Z¨ rich, 6, 135, 143, 151 u capital asset pricing model (CAPM), 15, 68, 71, 121, 122, 123 cash ﬂow statements, analysis, 122–123 chaos theory, 15 clothing business, turn-around situation, 127–129 company’s value, 62 conﬂict between bank and borrowers, 125–126 continuous-time ﬁnance, 15–16 continuous-time framework, 15 corrected credit shortfall risk, 81–82, 85 correction factor, 38, 45, 69, 70 country risk, 14 covered loans, shortfall risk of, 91–101 borrower vs collateral correlation, 93–97, 143 derivation of correlation, 93–94 maximum value, 95–97 minimum value, 94–95 example, 99–101 option theory approach, 91–93 vs uncovered loans, 92, 98 credit balance assets, 54 credit limits, exceeding, 40 credit risk, 14, 16, 124 credit shortfall risk, 64, 66 inﬂuence of individual parameters, 71–77 Credit Suisse Group, 6, 7 credit-worthiness, 5 credit-worthiness key ﬁgure, deﬁning, 21 credit-worthiness risk, 9 current account credit, 26–40 current account credit write-off risk hedging rate, 37 current account loans, 52 customer considerations, 145 data limitations, 12 death, risk of, 54, 71 debt rate, 65, 67–68, 121 discount rate, 71, 123–124 discounted free cash ﬂows, 121 discretionary income, 71 earnings, declining, companies with, 114 earnings statements, analysis, 122

45 equity. 108–10 exchange rate risk. 3 loan assessment. 139. 121 liquidity risks. 9. 14 modern credit risk analysis. 28–29 maximum debt. 15 market risks. determination. 140 interest rates acceptability of. specialist. 36 without repayments. 71 privileged salary claims in case of bankruptcy. 116–117 shortfall risk. 14 mortgage. 8 loan interest rate model. 13–17 classical methods. 60–63 limits to application. 51 with regular repayments. 71 liquidation value. 29 loan interest rate. 54 property price index. 27–28 proﬁt contribution rate. 112 risk. 69. 103–105. 36 ﬁxed advance with complete repayment. 54 marginal interest rate. 59–60 legal interest rate cap. for very competitive markets. example. 13. acceptability of. 105–106 loan commissions. 8 loan exposure models. 113 return on assets. 82–84 organisational measures. 107–108 three loans. 31 probability model. 14 option theory approach. 141–142 loan business. 11. 114. 54 IT support. 7–8 loan combinations. 23 need for. 40–41. 141–142 portfolio of securities. 143 jump-diffusion model. general case. 11 neural networks. 5. case study. 80–82 probability. 6 machine tools company. 45 Monte Carlo simulations. 14 operational risks. 3. 115 market price of risk. problems in. 143 private debtors. 115 maximum return on equity. 12 market risk premia. 114–115 loan terms. 83 KMV Corporation approach. 48 minimum loan interest rate. 14. 142–143 ﬁnancing cost rate. 108 several loans. 47–48 approximate solution. 16 market value. 9. 44 mean shortfall risk. 45 effective proﬁt contribution rate. 68 gamma function values. 14–16 loan granted indeﬁnitely. 99–100 net present value of loan transaction. 14 invalidity. 143 business 1985–99. 60–61 maturity transformation. 129–130 management agreement. 43. 4 basic formulae. return on. 151 rating system ampliﬁcation. 48–50 free cash ﬂow. 143–144 partially covered loans. 108 personnel. 39 investment income. 27–31 probability of cash ﬂows being fulﬁlled. 8–9 loan market value. 3. 108–112 maximum shortfall risk covered. 54 inventory limits. 108–110 effective loan interest rate. 33–35 ﬁxed interest loan without repayments. 135–138 combination of loans. 50–51 with partial repayments. risk of. 33 . 45 effective shortfall risk hedging rate. 70 insurance premiums. 21–23 repayment. 135–136. 43–44. 8 ﬁnancing rate. 107 two loans. 79–80 loan assessment ofﬁces. 19 in terms of ﬁgures. 13. 14 ﬁnancing companies. loans against. 42. deﬁnition. 36–37 property assets.168 Index loan supervision. 124 losses on loans. 103–117 partially covered loans. 15 Market Risk Amendment.

67–69. 107 transition matrices. 130–131 comparison of bank/model assessment. 85–87 company with a poor ﬁnancial year. 3. 71 validation. 124 determination of discount rate. 122 conﬂict between bank and borrowers. 31 deﬁning in terms of ﬁgures. 127–129 two loans. 15 stockmarket risk. 47–52 standard rate of interest. 70 winding up ofﬁces. 98 inﬂuence of individual parameters on. 121 determination. 123–124 risk premium rate. shortfall risk on. 4 shortfall risk hedging rate. 70 of the debts. model for calculating. 33–46 in general case of variable shortfall risk. model. 108–110 return rate. 54. 55 company with continuous business development. 131 starting position. 125–126 determination of credit risk. 62 variable shortfall risk. 53–55. rate of. 10–11 stochastic calculus. 92. agreed. 20–21 deﬁnition. 125 risk insurance premium. 71–77 private clients. 66–67 of assets. derivation of. 9–11 salaried employment. 134 limitation of loan risk. 139–141 three loans. model. 59–89 companies. 142 zero bond approach.Index return on equity. 124 overall view of procedure. 123–124 determination of relevant loan terms. 84 risk-adjusted values. income/unearned income from. 54 sensitivity testing. 98 unemployment beneﬁt. 12–13 tests of hypotheses. 6 term of loan. 124 determination of volatility. 121 testing. 54 salary claims in case of bankruptcy. 129–130 survival chance. 130–135 banks’ loan decision. 71 unemployment risk. 105–106 169 uncovered loans. 63–67 risk assessment for lending to a company analysis of cash ﬂow statements. 130 shortfall risk. 87–89 covered/uncovered loans to same borrower. 29–31 successor company provision. 14 success chance. 142 self-employment. 121–122 prudence in case of new loans/borrowers. income from. 31 Swiss Bank Corporation. 38–39 risk surcharge. 140 risk measurement. 13 ship mortgages. 123 analysis of earnings statements. 124 whole option theory. 14 turn-around situation. 131 loan risk determination according to model. 9. 134–135 loan risk assessment by banks. 47–52 volatility. 69–70 annual. 12 value of a company. 108–110 risk-adjusted loan interest rate per annum. 66 Compiled by Annette Musker . 53–54 vs covered loans. 131–134 model information. 80–82 sales staff. 122–123 analysis of balance sheets.

- APJFS Paper Docx Format Ver22_final Version
- Implementing Derivatives Models
- Financial Management
- UNIT - 3
- Portfolio Optimization Subject to Restrictions on Tracking-Error Volatility
- Final Doc Subbu
- How Standard Deviation Works
- Understanding BCAR for Property/Casualty Insurers
- Bloomberg Bval 2047437
- 013_200CreditPortfolioModels
- DEVELOPING PREDICTION MODEL OF LOAN RISK IN BANKS USING DATA MINING
- The Investigating of the Relation between Working Capital and Performance with Considering Financial Constraints in TSE
- Brm. Project
- FIN6102_Assignment_2__Jan_29 (FINAL)
- MPRA Paper 28672
- Soln Ch 27 Active Mgt
- impact of divident announce
- CRISIL Mf Ranking Methodology Sep 11
- Financial Valuation Analyst Certification
- A_fuzzy
- Economics 13 Chapetr
- 3
- Improving Returns from the Markowitz Model using GA- AnEmpirical Validation on the BSE
- Accounting Principles
- KPI_by Duncan Williamson
- Quarterly Crowding Report_2013Q3
- International Financial Management
- Capital Budget 2
- Chapter 6 - Modelling Innovation - Teaser
- 10 Risk and Uncertainty~INTRODUCTION TO FINANCE FOR SEGi DIA, DIBA, DIM

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue listening from where you left off, or restart the preview.

scribd