Você está na página 1de 96

Journal

The Capco Institute Journal of Financial Transformation

Recipient of the Apex Awards for Publication Excellence 2002-2013

Zicklin-Capco Institute Paper Series in Applied Finance

#38

09.2013

PUTTING IT ALL
TOGETHER FOR
OUR CLIENTS ...
Thats what makes FIS the global leader.

What does it take to be the global leader in


banking and payments technology?
It begins with the vision to assemble the best
suite of solutions and services in the industry.
It continues with the expertise to put all the
pieces together seamlessly. And its sustained
with the commitment to drive results for
our clients.
Theres nothing puzzling about our rise
to the top of the financial services game.
For more than 40 years, we have designed
and built technology that enables the realtime movement of money for our clients to
benefit their customers. And our more than
32,000 industry experts around the globe are
passionately focused on helping our clients
achieve new levels of success.

FINANCIAL SOLUTIONS PAYMENT SOLUTIONS BUSINESS SOLUTIONS TECHNOLOGY SERVICES

www.fisglobal.com
2012 FIS and/or its subsidiaries. All Rights Reserved.

Journal
The Capco Institute Journal of Financial Transformation

Recipient of the Apex Awards for Publication Excellence 2002-2013

Editor
Prof. Damiano Brigo, Director of the Capco Institute and Head of the Mathematical
Finance Research Group, Imperial College, London

Head of the Advisory Board


Dr. Peter Leukert, Head of Strategy for Global Financial Institutions, FIS, and Head
of the Capco Institute

Advisory Editors
Cornel Bender, Partner, Capco
Nick Jackson, Partner, Capco

Editorial Board
Franklin Allen, Nippon Life Professor of Finance, The Wharton School,
University of Pennsylvania
Joe Anastasio, Partner, Capco
Philippe dArvisenet, Group Chief Economist, BNP Paribas
Rudi Bogni, former Chief Executive Officer, UBS Private Banking
Bruno Bonati, Strategic Consultant, Bruno Bonati Consulting
David Clark, NED on the board of financial institutions and a former senior
advisor to the FSA
Gry Daeninck, former CEO, Robeco
Stephen C. Daffron, former Global Head of Operations, Morgan Stanley
Douglas W. Diamond, Merton H. Miller Distinguished Service Professor of Finance,
Graduate School of Business, University of Chicago
Elroy Dimson, Professor Emeritus, London Business School
Nicholas Economides, Professor of Economics, Leonard N. Stern School of
Business, New York University
Michael Enthoven, Former Chief Executive Officer, NIBC Bank N.V.
Jos Luis Escriv, Group Chief Economist, Grupo BBVA
George Feiger, Executive Vice President and Head of Wealth Management,
Zions Bancorporation
Gregorio de Felice, Group Chief Economist, Banca Intesa
Hans Geiger, Professor of Banking, Swiss Banking Institute, University of Zurich
Peter Gomber, Full Professor, Chair of e-Finance, Goethe University Frankfurt
Wilfried Hauck, Chief Executive Officer, Allianz Dresdner Asset Management
International GmbH
Pierre Hillion, de Picciotto Chaired Professor of Alternative Investments and
Shell Professor of Finance, INSEAD
Thomas Kloet, Chief Executive Officer, TMX Group Inc.
Mitchel Lenson, former Group Head of IT and Operations, Deutsche Bank Group
Donald A. Marchand, Professor of Strategy and Information Management,
IMD and Chairman and President of enterpriseIQ
Colin Mayer, Peter Moores Dean, Sad Business School, Oxford University
Steve Perry, Executive Vice President, Visa Europe
Derek Sach, Head of Global Restructuring, The Royal Bank of Scotland
ManMohan S. Sodhi, Professor in Operations & Supply Chain Management,
Cass Business School, City University London
John Taysom, Founder & Joint CEO, The Reuters Greenhouse Fund
Graham Vickery, Head of Information Economy Unit, OECD

Zicklin-Capco
Institute Paper Series
in Applied Finance
Ideas at Work

Cutting Edge

67

Sovereign Credit Risk in a Hidden Markov RegimeSwitching Framework. Part 2


Louise Potgieter, Gianluca Fusai

83

Viewing Risk Measures as Information


Wayne Tarrant, Dominique Gugan

A Journey to Operational Excellence


Julien Blanchet, Stphane Arvor, Bethsabe Fitoussi,
Galle Laboureur, Emmanuelle Mayet-Delors

High-level Debate
17

Enhancing Islamic Finance through Risk


Benchmarking
Rodrigo Zepeda

35

Credit Valuation Adjustment: The Devil is in the


Detail
Sylvain Prado, Bhavdeep Virdee

41

Aligning Marketing and Finance with Accepted


Standards for Valuing Brands
James Gregory, Michael Moore

47

CCP Regulations: Posing Challenges for Both


CCPs and Regulators
Jennifer Liu

55

The US Home Foreclosures and Federal Funds


Rate: Is the Channel Effect Muted?
Gke Soydemir, Andres Bello, Tzu-Man Huang

Dear Reader,
Time flies. It only seems like yesterday

industry, we aim to uncover the real is-

that I was writing the introduction to is-

sues and the prominent solutions avail-

sue 37 of the Journal. So what trends

able to financial markets participants.

have dominated the banking industry

and transparency, while reducing costs.


In the final phase of the debate, we examine the role of industrialization. No

since April? You do not have to look too

The three main areas for conference

longer synonymous with outsourcing

far to see that opinion is sharply divided.

discussion mirror the journey of recent

and cost-cutting, our panel will look at

On the one hand we have witnessed the

months while trying to establish the criti-

how industrialized institutions can help

return to healthy profits with many pun-

cal topics that will set the agenda to the

banks regain healthy levels of return on

dits arguing that banking has turned the

end of this year and beyond. We first ex-

equity, address overcapacity and emu-

corner and that we are seeing the first

amine the perfect storm of regulation,

late digital businesses in the delivery of

signs of sustained recovery since 2008.

risk, reporting and stress testing domi-

personalized products and services to

nating operations in banks. Here our

customers, including the latest genera-

On the other, there are those pointing

panel of experts tackles the persisting

tion of digital natives.

out that in spite of resurgent balance

challenges of credit, liquidity and lever-

sheets, we are still in the middle of a fun-

age risk. We also examine the ongoing

In the current issue of the Journal we

damental transformation of the sector,

challenges of regulation and the immi-

also give leading voices the opportunity

where complex regulation is expanding

nent impact of CCPs.

to present a range of views on many

with little or no end in sight; productivity

of these subjects. Whether you are at-

remains below benchmarks set by lead-

Having considered the bigger picture,

tending the conference or are one of

ers in other industries; and where inves-

we then focus on the growing role of

our readers I hope the articles help shed

tors remain skeptical although some

technology, in particular electronic trad-

new light on these topics and provide

optimism is returning.

ing. In the wake of the Nasdaq network

you with real-life insights that can be ap-

failure in August, we face up to the com-

plied to your own business.

All these themes are the subject of both

plexity that electronic trading now brings

this Journal issue and of the second

to the marketplace and weigh up the

Zicklin-Capco Conference, Industrial-

consequences for banks and investors.

ization and Innovation, A New Perspec-

We also consider the impact of high fre-

tive, in New York on October 23. Given

quency trading and ask whether it holds

Rob Heyvaert

the lively debate about the future of the

true to its promise to increase fairness

Founder, Chairman and CEO, Capco

Expanding Finance:
Conditions at the Boundary
At the present time, when we are still try-

topic, looking in particular at the Basel

This issue concludes with the Cutting

ing to understand how to move beyond

simplified regulatory CVA formula and

Edge section. In the first article we con-

the economic crisis, it may seem strange

the potential consequences CVA will

clude the investigation into how one

to talk about financial expansion. Yet, in

have for banks.

could expand sovereign credit risk mod-

a complex system, expansion may still

els by including regimes. Louise Potgiet-

occur. To understand why we need to

Part of the transformation and expansion

er and Gianluca Fusai present the practi-

take into account what is happening at

that is affecting the financial system as a

cal application of the model (developed

the boundaries of the industry.

whole and will have a heavy impact on

in the first part of the article, published

CVA is the full onset of CCPs. CCPs are

in issue 37). In the second article of the

Our Ideas at Work section for this issue

considered from a high-level perspective

cutting edge section, Wayne Tarrant and

deals with operational excellence, a topic

in Jennifer Lius article, where a discus-

Dominique Gugan suggest that a lack of

that ties in directly with financial transfor-

sion on the challenges for regulating and

sufficient information in risk management

mation and evolution. As Julien Blanchet

managing CCPs is presented.

can lead to ambiguous risk situations.

and coauthors explain, financial institu-

The need for the reporting of multiple risk

tions are now targeting operational excel-

Our journey into expansion continues

measures or, ideally, of the whole loss

lence and redefining their business model

with marketing. Marketing is an area that

distribution is advocated. An expansion

using methodologies such as Lean Six

seems to be more qualitative than quan-

of current regulation in terms of multiple

Sigma well established in other indus-

titative, in that measuring direct market-

risk measures is also suggested.

tries. This is just one way that institutions

ing effectiveness and value may prove

are propelling themselves towards a more

difficult. James Gregory and Michael

This concludes this issues journey into

efficient and industrialized regime.

Moore challenge this view and argue that

the expansion of finance and conditions

there is the potential to expand market-

at its boundary. But of course the journey

Our High-level Debate section looks at

ing metrics by aligning marketing and

never really ends. We hope our readers

the expansion of financial boundaries by

finance with accepted standards for

will accompany us as we explore the

examining the opportunities for Islamic

valuing brands.

possibilities at the limits of the financial

finance at a time of rapid change, espe-

industry and of course we also continue

cially by developing new points of view

The High-level Debate section concludes

for risk management. Rodrigo Zepedas

with a discussion of the US house mar-

article tackles this theme.

ket and, in particular, whether the Fed-

to welcome submissions.

eral Funds Rate has a direct impact on


On a different note, but still related to

US home foreclosures. We know how

expansion and transformation, the prob-

important the house market has been at

lem of counterparty risk and credit valua-

the onset of the crisis back in 2007/2008,

tion adjustment (CVA) is also pushing the

and one of the key themes any expan-

boundaries of computational capabili-

sion of finance has to consider is how

ties, computer science solutions, archi-

governments and central banks may try

tecture, regulation and capital strategy.

and manage the house market. Gke

I have extensively researched this area

Soydemir, Andres Bello and Tzu-Man

Prof. Damiano Brigo, Head of the Capco

myself in the past 10 years. The article

Huang take up this challenge by investi-

Institute and Head of the Mathematical

by Sylvain Prado and Bhavdeep Virdee

gating the relationship between the Fed-

Finance Research Group, Imperial

makes an interesting contribution to this

eral Funds Rate and foreclosures.

College, London

On behalf of the board of editors,

Ideas
at
Work

Ideas
at
Work

A Journey to
Operational Excellence
Julien Blanchet, Managing Principal, Capco
Stphane Arvor, Principal Consultant, Capco
Bethsabe Fitoussi, Principal Consultant, Capco
Galle Laboureur, Senior Consultant, Capco
Emmanuelle Mayet-Delors, Consultant, Capco

Introduction

plans currently carried out by banks.

Lean Six Sigma was first used in banking in

layers of questionable controls to existing


ones. Today, the efficiency of controls needs
to be measured and challenged.

operational efficiency program for all its enti-

The four operational challenges of the


financial industry

ties, including GE Capital. The methodology

After years of uninhibited growth where expan-

Improve performance management and

proved successful very quickly and spread to

sion took place without rationalization con-

monitoring. Few financial institutions possess

the whole financial industry, as an attempt to

cerns, many in the financial industry have con-

relevant and efficient measurement systems

save money in the majority of cases.

cluded that the 2008 crisis marks the end of an

of their operational performance. Even though

1995, when General Electric launched an

era. Nowadays, competition is harder, margins

a significant effort is spent on producing and

Today, banks have mastered most operational

are smaller and risks are monitored more close-

aggregating indicators at each level of the or-

efficiency techniques including Lean Man-

ly. Other factors including diminishing liquidity

ganization, managerial decisions are too often

agement, Lean Six Sigma, Kaizen and BPM.

and tighter regulatory constraints are forcing

based on more subjective criteria.

Thousands of employees have been trained

major players in the banking industry to mul-

and certified. Determined to tackle specific

tiply their efforts and implement new solutions.

challenges, individual banks have developed

Transform management culture. Employees at large banks are impacted by the rapid

their own dedicated transformation programs

In this context, four major operational chal-

changes in the financial industry and they are

and operational efficiency centers of expertise.

lenges need to be addressed:

the ones who live transformation on a day-to-

This awareness of the need for transformation

Increase the operating models efficiency.

needs to shift from technical expertise to team

was reinforced after the 2008 crisis, but when

Flawless execution is a must. Institutions need

leadership. While they are asked to implement

the liquidity crisis broke in 2011, the drive for

to focus on the delivery of services to custom-

top management decisions, they are also ex-

Operational Efficiency became even more ur-

ers, develop best practice, and share and im-

pected to create a work environment where

gent. Financial institutions are now targeting

prove financial performance.

people can develop their talents and perform

day basis. As a result, the role of managers

operational excellence and redefining their

to the best of their abilities.

business model with a strong focus on client

Secure risk management. While operational

satisfaction, performance, and cost drivers. As

risks have increasingly stolen the limelight

Responding to these challenges and getting

a result, operational efficiency programs have

over the past decade, the response of financial

optimum results requires strong convictions

become key components of current adaptation

institutions has too often consisted in adding

and rigorous methodologies.

The Capco Institute Journal of Financial Transformation


A Journey to Operational Excellence

Conviction is everything

Methodologies

the transfer of knowledge and training to the

Whichever way you look at it, operational

In this section we focus on three combined

next generation. Here, knowledge transfer in-

excellence can be reached anytime and any-

basic principles to reach operational excel-

volves a train-the-trainer model of sustainable

where.

lence: the way of doing, the way of thinking

operational excellence. This means that knowl-

and the way of growing.

edge is maintained and re-used and banks can

Operational excellence is a decision

develop their own expertise, become indepen-

Excellence is not spontaneous. Current pro-

Basic principles

dent in the design and management of their op-

cesses and organizational schemes at finan-

Doing

erational excellence efforts, and foster a culture

cial institutions are not driven by efficiency

After a problem has been identified, a rigorous

of continuous improvement.

alone. Unchallenged historical decisions and

problem-solving process is applied. The key

constraints also lead to complexity and waste.

to solving a problem is to treat its root cause

In short, success hinges on how well each em-

rather than the symptoms, in order to avoid

ployee at every level of the organization helps

continuous fire fighting.

drive an evolutionary learning dynamic in the

Organizations demonstrating higher operational efficiency rates do not achieve this by

company culture.

chance. All of them have a dedicated program

Depending on the scope and project objec-

in place, proving that excellence is a conscious

tives, different problem-solving approaches,

Methodologies and toolkits

decision. This awareness is necessary at each

drawing on a large toolkit, can be applied to

Operational efficiency programs and projects

level of the organization, from the highest

solve problems and achieve high performance.

often set similar objectives:

management levels to operational staff.

Greater efficiency and reduced costs;

Thinking

Decreased operational risks;

Operational excellence requires


knowledge as well as experience

In order to meet each clients specific needs,

Performance management;

we propose a building-block approach based

Cultural change.

Excellence is not straightforward. It requires

on three pillars: processes, organization and

the right balance between technical knowledge

people.

and experience. It takes research and out-ofthe-box thinking to define new practices and

More specifically, quantitative gains will be derived from:

Focusing on processes generates value-

Direct budget-impact productivity;

new ways of working. This knowledge is avail-

stream optimization, reduces process vari-

Reallocated productivity gains;

able on the market and various methodology

ability, eliminates non-value adding activity

Cost avoidance and cash reduction;

and increases performance monitoring.

Increase in turnover;

Focusing on organization increases the

Reduced operational risks by cutting the

certifications, including Green and Black Belts


and Lean experts can help measure and rec-

ognize the relevant level of expertise needed

global performance of a business and per-

to meet your organizations goals. Above all,

mits the design of new operating models.

excellence cannot be reached in an unstruc-

number of operational incidents.

Focusing on people involves implementing

The list of methodologies available on the mar-

tured way. Change is a process that follows

leadership through powerful communica-

ket to improve operational efficiency is very

phased approaches.

tion, clear, and transparent guidelines, the

long. The purpose of this article is to focus on

development of interpersonal skills, and

the most widespread with the most impressive

a culture of mutual trust based on shared

track records: Lean Six Sigma and Lean Man-

vision and values.

agement.

Operational excellence is a continuous


effort

Even when the will to succeed exists there


is always the risk that continuous improve-

These three modules are complementary and

Lean Six Sigma

ment momentum will slow down. Operational

can be applied either on a stand-alone basis

Lean Six Sigma is a continuous improvement

efficiency expertise is also difficult to retain

or simultaneously. Tailoring the panel of ap-

methodology combining the two most power-

and renew. Senior management can lose fo-

proaches to fit specific problems increases

ful and popular quality trends of recent times.

cus once they believe that performance has

the efficiency of the solutions, contributing to

While the Lean philosophy concentrates on

reached an acceptable level. So you con-

higher performance.

reducing lead times and optimizing costs, Six

stantly need to re-invent and learn in order to


guarantee continuous improvement.

Sigma is a data-driven approach focusing on


Growing

quality targets critical for customers and re-

One of Capcos main focuses is to encourage

ducing the variation in process outputs.

Ideas
at
Work

Lean

Six Sigma

Focuses on process
to eliminate non-value
added activities and
improve velocity

Focuses on variance
to remove defects and
improve quality and
reliability

Provides tools to
analyze flow and delay
time

Provides statistical
techniques and other
tools to identify root
causes of defects

Lean Management

Our success story

Lean Management projects aim at applying


the lean principles to the way teams work to-

Setting up an operational efficiency


program at a leading French bank

gether. This in turn leads to a continuous im-

At the time Capco started to propose the use

provement culture that applies to the whole

of the Lean Six Sigma methodology to drive

organization. At the end of the journey, each

the operational excellence of a leading French

individual will not only carry out his or her role,

bank, considerable success had been docu-

but propose and implement ways to improve it

mented in the manufacturing sector, but little

on a daily basis.

had been done in service-based industries.


The first step in our deployment was to iden-

Figure 1 Lean Six Sigma

The methodology is based on an I do, we

tify the business leaders and convince them

do, you do principle. For each implemented

to use the methodology by showing what they

tool, managers and/or staff will be trained then

could expect to achieve.

coached on the way the tool has to be used.


A dedicated team of experts is specifically set

They will then have the ability to adapt the

The mobilization of the organization

up for a pre-defined period of time to achieve

tools to their own situation.

around a common set of objectives

specific and measurable objectives. This project team comprises:

Four main objectives were identified by the


As with Lean Six Sigma, a dedicated team of

client:

Certified staff (Green and Black belts);

experts is specifically created for a pre-defined

Operational experts (from the process

period of time to achieve specific and mea-

where the project is launched), trained in

surable objectives. The team will conduct a

the Lean Six Sigma methodology and can-

phased approach (mobilize, diagnose, plan,

didates for certification.

implement) in order to:

The team will conduct a phased approach


(DMAIC1, DMADV2, PDCA3, etc.) focusing on

Make sure operational staff trust the axes

facts that help bring all stakeholders to a com-

for improvement, then apply continuous

mon agreement on the diagnosis and on the

improvement mechanisms.

axes for improvement to achieve the objectives.

process efficiency;

Decrease operational losses by reinforcing


operational risks mitigation;

Implement foundations for continuous


improvement;

Decrease operating costs by improving

Improve business decision-making through


the implementation of indicators;

Spread a culture of continuous improvement.

The pilot
We agreed to deploy a six-month Prove the

The strength of the method lies in its power

Concept pilot phase, led by one Master Black

to change peoples behaviors. The methodol-

Belt and two Black Belt Capco resources, cov-

The strength of the methodology lies in the

ogy tracks process inefficiencies but also im-

ering four parallel projects, in order to show

power of data. Facts and figures are powerful

proves the ability to meet customer demands

the impact, validate the methodology, and tai-

tools that bring objectivity. In addition, opera-

by adapting and enhancing the teams capac-

lor it to our clients specific requirements.

tional experts participating in the project free

ity and competencies.

up a significant amount of their time to chal-

The pilot projects generated a return on in-

lenge their own process, build the diagnosis,

The toolkit is complete and tackles process,

vestment which exceeded expectations. They

and seize the opportunity to improve it.

organization, performance, and mindset im-

yielded significant cash flow improvement and

provement. Moreover, the methodology in-

cost savings, and at the same time provided

When deploying this methodology it is im-

volves all layers in an organization, ensuring

better quality, as well as higher staff and client

portant to keep the momentum alive. The

sustainability.

satisfaction.

implementation of the middle- and long-term

10

solutions (after the project team intervention)

When deploying the Lean Management meth-

requires an effort which is often underestimat-

odology it is essential to get a high level of

ed. In addition, Lean Six Sigma tools are less

commitment from managers and operational

well adapted to inefficiencies, such as mindset

staff in order to implement the continuous im-

and organizational issues that are not related

provement mechanisms.

to core process.

1 Define, Measure, Analyze, Improve, Control


2 Define, Measure, Analyze, Design, Verify
3 Plan, Do, Check, Act

The Capco Institute Journal of Financial Transformation


A Journey to Operational Excellence

one internal program manager, and a steering

Examples of quantitative and qualitative gains of the pilot:

committee of key stakeholders overseeing all


the aspects of the roll-out (top management,

Quantitative gains:

heads of business lines and support functions,

Significant RWA reduction through better coordination between risk and finance on the

finance coordinator, and Capco delivery man-

Basel II process

ager).

12% of productivity gains in the middle office


Discuss the right topic at the right level

Qualitative gains
The certification of seven Green Belts and the mobilization of numerous stakeholders

The governance structure starting with lead-

around the methodology and tools

ership commitment and ending with proj-

A culture change in project management methodology and approach

ect-based improvement activities ensured

The reinforcement of process reengineering in the search for operational excellence

harmonious top-down and bottom-up com-

The definition of sustainable indicators to monitor processes

munications. This also ensured that follow up

The introduction of change management in project implementation

of the daily performance of the business pro-

cesses took place at the right level. With proThis success story led to the creation of an operational efficiency program sponsored by the

ductive meetings and review sessions sched-

top management and covering all the banks entities.

uled on a regular basis, the program was able


to create and sustain its momentum.

Building up capacity

Top management support

Develop a project portfolio roadmap

Mobilize a dedicated expert team

Strong sponsorship from top management

Determining which projects to select for the

After the success of the pilot phase, other in-

is key to communicating the vision and en-

best and fastest returns was a real chal-

ternal and Capco experienced resources were

suring people commitment. The deployment

lenge for the client at the beginning of the

added to the program team which conducted

governance framework of the program was

Lean Six Sigma program. To address this

Lean Six Sigma projects throughout the bank.

established jointly by one Capco consultant,

need for fast and high return-on-investment,

Ensure client autonomy through mixed


project teams: training and certification of
operational resources
In order to ensure client autonomy and op-

Profile and certification


Is a senior manager
Has no specific technical LSS
knowledge
Not an LSS certification

Lean Six Sigma structure

Champion

Role
Coordinates program
Links sponsors and senior
management
Reports on program progress

erational teams commitment, each Lean Six


Sigma project team was organized around
mixed resources including program team
members and operational key contacts of the

Is a senior project manager


Follows the Master Black Belt
certification process

Master
Black Belt

May lead a stream and/or support


program manager
Coaches Black Belts
Certifies Green and Black Belts

specific area where the project was led. During


the project, appropriate training and coaching were provided so that they had complete
control of tools and the methodology. This
also helped promote a culture of continuous
improvement. Each participant was Yellow or
Green Belt certified which provided a valuable
credential demonstrating the participants new

Are focused on projects (no


operational responsibilities)
Are fully dedicated to LSS program
Follow the Black Belt certification
process
Are in the project team
Promote LSS internally
Follow the full Green Belt training
and certification process

Black Belts

Define project approach


Manage and coach Green and
Yellow Belts
Are responsible for key deliverables

Green Belts

Take ownership of key deliverables


Participate to the whole phased
approach

Yellow Belts

Participate in key phases of the


project
May attend sponsor meetings

knowledge and experience.

Establish strong program governance


An operational efficiency program depends on

Are process owners


Follow a 1-day training to understand
LSS key concepts
Not an LSS certification

well thought-out and executed deployment


governance.

Figure 2 The various levels of seniority for the LSS methodology

11

Ideas
at
Work

sought a clear diagnosis. At the same time,

Key lessons learned when implementing an operational efficiency program

an operational risk analysis (FMEA4) was conducted to identify high-risk process steps and

Strong commitment from top management

Program objectives are set up

Shop floor visits (Go and see)

Role modeling (Walk the talk)

estimate risk associated with specific failure


causes.
The Improve phase resulted in the implemen-

Alignment of program with the banks strategic projects

tation of suitable and sustainable changes,

Transparency in communication (including efficiency gains)

mainly focused on:

At all organization levels (including staff communication)

For each program key milestone

non-value added tasks;

Strong encouragement of the team leaders and operational staff to participate in the

process analysis and the research of improvement solutions

Reducing process time cycle by removing


Improving polyvalence across team members to enable them to better react and
adapt to flows;

Clear reward process based on results

Efficiency gains realization and accounting

Guaranteed through a strong financial accounting process

HR involvement where there are organizational consequences

Present program to social partners in advance

Decreasing data providers bills;

Reducing operational risk by setting up


standardized procedures and set in stone
production cut-off.

Management tools including capacity model,


a comprehensive process cartography was

was clarified and the scope defined. Via VOC

operational risk dashboard, and data quality

designed to feed the programs projects road-

(voice of customer) and VOP (voice of process)

follow-up dashboard were set up to help the

map. Fifty-four macro-processes covering the

interviews, the main pain points in the pro-

management pilot its activities and monitor cli-

whole bank were identified and then prioritized

cesses that were critical to quality were identi-

ent delivery quality.

by the top management based on the program

fied. The mapping of the process highlighted

objectives and expected benefits.

non-value added steps and unmitigated op-

With these indicators, the improved processes

erational risks, while customers raised data

were monitored during a Control phase during

As the projects were delivered and a positive

quality issues and production delays. At the

which productivity gains of 13% of the initial

image of the program spread across the bank,

end of this phase, a business case was calcu-

baseline and cash savings of 500,000 were

the number of participating departments in-

lated based on current cost and target gains

realized.

creased steadily over the whole bank.

objectives.

Lean Management project


DMAIC project

During the Measure phase, data tied to the

The organization of Lean Management proj-

The following example illustrates an applica-

speed and quality of the instrument creation

ects depends on the maturity of the business

tion of Lean Six Sigma DMAIC methodology in

and modification process were collected.

areas where it is to be applied.

a market data administration process.

The pain points identified during the previous


phase were analyzed, ensuring that stakehold-

At the start of a program it usually includes

The so-called DMAIC phase started with the

ers reached a common agreement concerning

three phases:

Definition of the problem: due to the expan-

the existence, size, and impact of the problem.


The Diagnosis phase aims at building aware-

sion of its geographical perimeter, the operational team was facing an increasing number

To ensure data quality, operational staff were

ness in the management team about how

of instrument creation and modification re-

involved during the measure campaign.

things are running in their areas of responsibil-

quests with no additional staff. How could pro-

12

ity. Pain points as well as positive points are

cess efficiency be improved without impacting

Next, the Analyze phase was performed and

observed and shared with the whole team in

process capability and operational risk?

priority was given to root causes identification

order to create a sense of urgency: We must

and prioritization of pain points. In this phase,

change.

To identify what the process entailed, a project

problem solving techniques (e.g., Ishikawa

charter was created. The problem statement

and 5 Whys) involved every key player as they

4 Failure Mode and Effects Analysis

The Capco Institute Journal of Financial Transformation


A Journey to Operational Excellence

The Vision phase will then define precisely

all conditions are met, the balanced combina-

where the team wants to be in the mid-term

tion of governance and high quality delivery

and the path to get there: This is what we

will move your organization to higher perfor-

want to be in one year.

mance levels, whatever your current situation,


and initiate the continuous improvement mo-

The most important activity in these two

mentum towards excellence.

phases is the alignment of the whole managing team concerning what needs to be done

This is neither an easy road, nor a complicated

and why.

one. It is a pragmatic approach where the difficulty does not lie in the concepts but in the

The Implementation phase involves coaching

execution. The challenge is for the bank to

managers on a daily basis in order to deploy

believe sufficiently in the benefits in order to

tools such as visual management, efficient

attempt and sustain the effort. The elegance of

performance dialogue at all levels, and struc-

our model is its simplicity. The challenge is in

tured problem solving methods. The main

its implementation. But operational excellence

challenge for the lean management team is to

will be achieved with common sense and rigor.

get the managers to understand the sense


of the tools so that they can not only apply
them but also adapt them in the future for new
projects.

Conclusion
Your journey to operational excellence will not
start unless you decide to take it. Banks believing that spontaneous momentum is all that
is required will never make it. But once you
take the decision you can kick off safe in the
knowledge that there are methodologies with
a proven track record in the financial industry.
In order to increase the chances of achieving
your objectives, your operational efficiency
program team will combine various methodologies and project approaches that make it
possible to adapt effectively to your specific
challenges and constraints. All methodologies
will address process efficiency and simplification, optimization of team management, and
continuous improvement. However, a methodology alone is not sufficient. The operational
efficiency team will need to combine it with
in-depth knowledge and experience of the
banks business processes.
Results are based on well-known tools but
their effective implementation depends on the
proper prioritization of opportunities and the
skills of the operational efficiency team. When

13

High-level Debate

Enhancing Islamic Finance through Risk Benchmarking


Credit Valuation Adjustment: The Devil is in the Detail
Aligning Marketing and Finance with Accepted Standards for Valuing
Brands
CCP Regulations: Posing Challenges for Both CCPs and Regulators
The US Home Foreclosures and Federal Funds Rate: Is the Channel
Effect Muted?

16

High-level Debate

Enhancing Islamic
Finance through
Risk Benchmarking
Rodrigo Zepeda Independent Consultant and Associate, The Chartered Institute for
Securities & Investment

Abstract

superior to Western interest-based counterparts because

In essence, Islamic finance comprises financial services and

they are more equitable, stable and ethical, and offer less

products that are compliant with the religious, moral and

risk and volatility. Consequently in light of these propositions

ethical ways of Shariah or Shariah law. Shariah is not a

this paper will therefore argue that the development of a new

comprehensively codified set of written laws and rules, but

specialist risk benchmarking framework for public and

rather it represents a diverse and dynamic body of divine

private Islamic BFIs operating regionally (excluding central

injunctions which are applied to every aspect of a practic-

banks), is the first and necessary step towards empirically

ing Muslims life. This means that Islamic finance services

proving this claim of Islamic superiority right. Furthermore

and products must ensure that they adhere to all aspects of

it is submitted that the development of a collaborative risk

Shariah, both in the way they are structured and implement-

benchmarking framework for Western and Islamic banks and

ed. While it is beyond doubt that Islamic finance is a globally

financial institutions would contribute to the enhancement

established method of financing, Islamic banks and financial

of Islamic finance by offering three significant advantages.

institutions still face a number of operational difficulties. For

These include: (i) contributing to enhancing the legitimacy of

instance, it has been argued that the Islamic finance industry

Islamic finance; (ii) enhancing risk management systems and

is under-regulated, it lacks global cohesion and uniformity,

practices in Islamic BFIs; and (iii) affording realistic oppor-

and owing to its somewhat restrictive nature it has difficul-

tunities for the offering of new Shariah compliant Islamic fi-

ties in developing new Shariah compliant products and ser-

nance products. This paper will therefore provide a high level

vices. Other arguments have been made that Islamic banks

analysis of the theoretical framework within which the new

and financial institutions offer methods of finance which are

Islamic finance risk benchmarking could operate in practice.

17

Introduction

An overview of Islamic finance

The most recent estimates point to global Islamic finance assets reaching

Islam is a monotheistic faith which forms the second largest religion in

approximately US$1.8 trillion by 2013, with the top 20 Islamic banks reg-

the world, with over 1.6 billion Muslim followers of Islam. For Muslims, Is-

istering a growth of 16% over the last three years, while holding over 50%

lam means submission to the will of Allah, who is believed to have mani-

of Islamic banking assets globally [Ernst & Young (2012)]. Figures from

fested himself via revelations made to the Prophet Muhammad (pbuh) in

2008 show a total of 614 registered Islamic finance institutions spanning

the 7th century CE. These recitations were revealed to the Prophet Mu-

across 47 countries, and while established Islamic finance centers such

hammad (pbuh) during his lifetime by the angel Gabriel (Jibril), and were

as Malaysia and the Middle East (Egypt, Iran, Kuwait, Qatar, Saudi Arabia)

written down by him in the Quran. In modern usage Islamic Shariah)

have taken center stage in terms of asset levels, there are new markets

refers to a code of law or divine injunctions that regulate the conduct

on the horizon such as Iraq and Indonesia [The Banker (2008)]. Added

of Muslims in their individual and collective lives, and which relate to a

to this, the Islamic finance industry has registered more than 20% an-

number of different branches such as Aqid (matters of belief and wor-

nual growth since 2000, and with Muslims accounting for approximately

ship); Akhlq (matters for disciplining oneself); Ahkam (socio-economic

20% of the global population, the potential for future growth worldwide is

and legal systems); Fridh (obligations); and Nawhi (prohibitions) [Ayub

enormous. Nevertheless, there are still a number of key challenges facing

(2008)]. In practice Shariah (or Shariah law) is a principles-based legal

banks and financial institutions (BFIs) operating within Islamic finance

system which is derived from both primary and secondary sources (Usul

going forwards. These include the profitability of Islamic banking continu-

al-Fiqh). The primary sources are the Quran (the divine scripture), the

ing to lag behind that of conventional Western banking in the same mar-

Sunnah (the teachings, practices, and traditions of the Prophet Muham-

kets, as well as sub-scale operations; incomplete market segmentation;

mad (pbuh)), and the Hadith (the acts or sayings ascribed to the Prophet

limited client engagement; absence of technologically-orientated value

Muhammad (pbuh) and written down). The secondary sources are for-

propositions; and a very basic operational risk culture [Ernst & Young

mulated through ijtihad (independent reasoning) which consists of the

(2012)]. In this new post-financial crisis era, it has been argued that firms

extraction, interpretation, and deduction of principles by qualified Mus-

need to implement a new risk paradigm, which includes clearer owner-

lim legal scholars or jurists (mufti), belonging to different Muslim schools

ship of risks, the development of a true risk culture, and improved trans-

of law or thought (madhhab). This interpretational process may take the

parency, understanding and modeling of risk [Gerken et al. (2010)]. Yet at

form of either ijma (principle reached via consensus of Muslim jurists or

the same time it has been observed that many Islamic BFIs based in the

scholars) or qiyas (principle reached via legal and philosophical deductive

Middle East (i.e., Bahrain, Lebanon, Saudi Arabia, Qatar, the United Arab

analogy). Moreover, according to Schoon (2008), The rules laid down in

Emirates (UAE)) lack risk management systems or software, and are lag-

shariaa govern every aspect of a Muslims life, including the way he con-

ging behind their Western counterparts in terms of effective risk manage-

ducts his business, the criteria for valid contracts and the prohibitions.

ment practices [Deloitte (2010), (2011)]. This paper will therefore argue

Therefore Shariah is to be seen not as a one-off choice, but rather a way

that the development of a new specialist risk benchmarking framework

of life for a practicing Muslim.

for public and private Islamic BFIs operating regionally (excluding central
banks) would contribute to the enhancement and development of Islamic

The term Islamic finance therefore refers to the provision of financial

finance by offering three significant advantages. These include: (i) con-

services, transactions, or contracts (aqd) which adhere to Shariah, or

tributing to enhancing the legitimacy of Islamic finance; (ii) enhancing risk

rather the rules making up Islamic commercial jurisprudence (fiqh al-mu-

management systems and practices in Islamic BFIs; and (iii) affording

amalat) [Jabbar 2009]. Under Islam all financial services, transactions,

realistic opportunities for the offering of new Shariah compliant Islamic

or contracts are valid or permissible (halal) unless they are either forbid-

finance products. This paper will therefore provide a high level analysis

den (haram), or fall under the purview of any one of the aforementioned

of the theoretical framework within which the new Islamic finance risk

Shariah prohibitions. In respect of the former, all products or services

benchmarking could operate in practice. The paper will first explain what

which involve inter alia pork (or pork products); alcohol; armaments;

Islamic finance is, as well as providing an overview of the main financial

blood (and blood by-products); carcinogenic drugs; pornography; or car-

products offered. Next the paper will discuss financial risk and financial

nivorous animals, are considered to be haram. In respect of the latter,

risk modeling within BFIs, as well as discussing what benchmarking is,

the Shariah prohibitions refer to prohibited contracts containing any of:

and how financial risk benchmarking may actually be undertaken in prac-

(i) riba (essentially interest); (ii) gharar (uncertainty); or (iii) maisir or qimr

tice. Finally, the paper will apply these ideas with a view to developing a

(gambling and speculation or games of chance). In essence riba means

theoretical high level roadmap for an Islamic finance risk benchmarking

excess, and it covers any ex ante predetermined fixed payment that is

framework for Islamic BFIs.

guaranteed where the rate of return is tied to the maturity and principal,
rather than the performance, of an investment [Mohammed and Kianfar

18

(2007)]. The prohibition of riba is based on a number of different verses

The Capco Institute Journal of Financial Transformation


Enhancing Islamic Finance through Risk Benchmarking

from the Quran which proscribe the use of riba.1 The principle is also

Murabaha (cost-plus or purchase and resale contract)

based on the fact that Islam does not recognize the intrinsic or time-

The Murabaha contract is essentially a Shariah compliant purchase and

value of money itself, only as a medium of exchange, and therefore pure

resale contract involving halal commodity or goods. At its simplest it may

returns based on money or loans are seen to be unfair or immoral. Islamic

involve the buyer identifying a commodity or goods (e.g., a car) which the

finance profits need to be accrued through the applied use of capital,

Islamic bank will purchase at the behest of the buyer, and then sell on to

land or labor, rather than through the lending of money, as well as deriv-

the buyer at a fixed rate of profit, i.e., cost-plus sale. The cost price of

ing from the application of risk-sharing mechanisms which encourage

the commodity is disclosed to the buyer and the profit element reflects

mutual co-operation (taawon).

the risk (mukhatarahof) of ownership and business risk which the Islamic
bank must take on prior to the re-sale of the commodity (as no increase in

The prohibition against gharar covers highly speculative or extremely risky

the mark-up is allowed if the buyer does not take up the goods). However,

transactions, and therefore contracts which involve such uncertainty or

the majority of Murabaha contracts are undertaken on a deferred pay-

ambiguity (e.g., uncertain or non-existent goods) are void under Shariah

ment basis through the use of a Shariah compliant credit sale (Muraba-

[Fadeel (2002)]. Essentially this means that Islamic financial contracts

ha-Bai Muajjal). Consequently, in a Murabaha and credit sale transaction

should not lack specificity of terms (e.g., sale price, deliverability, quantity,

the Islamic bank would purchase the commodity identified by the buyer,

quality), and open-ended or deliberately ambiguous terms are therefore

and would then re-sell the goods to the buyer on a cost-plus basis with

prohibited under Shariah [Jabbar (2009), 24]. The prohibition of maisir

repayments to be made via installments. This type of deferred payment

(or qimr) circumscribes the conditions in which wealth can be acquired

sale is Shariah compliant so long as the period provided for future repay-

via speculation (or games of chance).2 Extreme speculation in the form

ment is fixed beforehand, although it should be noted that some com-

of outright gambling such as roulette or spread-betting is prohibited, but

mon practices include the purchase and sale of goods that never move

since all commercial activities entail a certain degree of speculation, the

from the warehouse, e.g., construction materials. It has also been noted

distinction may be somewhat difficult to make in practice under different

that certain Islamic banks have specific space allocated to them in mer-

circumstances. Furthermore, in light of financial market turbulence follow-

chant warehouses where goods to be offered under murabaha are stored

ing on in the wake of the recent global financial crisis, commercial activities

or moved to demonstrate physical possession, not merely constructive

that may have previously been accepted by Islamic scholars as permis-

possession [Abdulkader et al. (2005), 64]. The emulation of traditional

sible may now instead be regarded as falling foul of the gharar or maisir

Western interest-based financing by the Murabaha means that in practice

prohibitions. Indeed Hanif (2008) argues that in principle it is a question

it has become very widely used for trade financing purposes, however

of the degree of speculation involved, and whether the intention behind

Islamic banks have to face additional asset risk, greater fiduciary risks,

the transaction is purely speculative, or whether it is intended to realize

greater legal risk and the Sharah compliance risk [Ayub (2008), 233].

a gain from some productive effort. However, it is invariably the case that
the degree of speculation and the subjective intention will be interpreted in

Mudaraba (trust or profit sharing contract)

different ways by different jurisdictions, and the interpretation of Shariah

The Mudaraba (or Mudarabah) contract is a type of joint venture between

rules by different Islamic Shariah Supervisory Boards (SSBs). Therefore,

two parties whereby one party, which is typically the Islamic bank, will act

while in theory the principles of gharar and maisir can be established in

as the investor (rabb almal or rabbul-ml), and the other party will provide

abstracto, in practice the principles may be applied in many different ways

the expertise or act as the entrepreneur (mudarib). It is essential for a true

according to relevant regional, jurisdictional, and SSB influences. For in-

Mudaraba that the investor has no right to participate in the management

stance, for some Islamic finance scholars it is believed that certain types

of the enterprise and that the entrepreneur provides no investment under

of equity investments in halal companies are accepted under Shariah,


whereas other types of investments in equity derivatives (e.g., index-linked
derivatives) are not acceptable, as they involve speculation on the movement of an equity index [Hanif (2008)]. While this may be an oversimplification, the example demonstrates the range of viewpoints that the principles
of gharar or maisir may encompass in practice. Nevertheless, in modern
Islamic finance practice, many Islamic finance contracts have now come
to be widely accepted as having an underlying structure which conforms
to all aspects of Shariah. Many of these nominate contracts are based on
variations of the underlying principle of bai (or sale), and a few of the most
widely used contracts (although far from exhaustive) will be discussed in
brief detail here.

1 For example, Surah al-Rum, 30 verse 39 states: That which you give as Riba to increase
the peoples wealth increases not with God; but that which you give in charity, seeking
the goodwill of God, multiplies manifold. Surah al-Nisa, 4 verse 161 states: And for their
taking Riba although it was forbidden for them, and their wrongful appropriation of other
peoples property. We have prepared for those among them who reject faith a grievous
punishment. Surah al-Baqarah, verse 275 states: Those who take Riba shall be raised
like those who have been driven to madness by the touch of the Devil; this is because
they say: Trade is just like interest while God has permitted trade and forbidden interest.
Hence those who have received the admonition from their Lord and desist, may keep
their previous gains, their case being entrusted to God; but those who revert, shall be the
inhabitants of the fire and abide therein forever [Ayub (2008), 45].
2 For example, Surah al-Maidah, 5 verse 90 states: O you who believe! Intoxicants and
gambling, sacrificing to stones, and divination by arrows, are abominable actions of Satan;
so abstain from them, that you may prosper. [Ayub (2008), 62].

19

the contract. Any profits that arise from the joint venture will be shared by

Sukuk (investment certificate contracts)

the parties in pre-allocated proportions, although it is only the investor that

Sukuk are medium (3-10 year) to long term (10+ years) Shariah com-

bears the risk of loss. If more than one investor is investing in the Mudaraba

pliant trust or investment certificates that are economically similar to

the investors are entitled to pre-allocate losses inter se. As compared to

Western (debt-finance) bonds. However, Western bonds pay fixed in-

traditional Western financing, the Mudaraba functions in much the same

come (interest) to bond-holders in the form of coupons, which therefore

way as equity financing ventures (e.g., seed financing), however there are

violates the prohibition against riba. Consequently, sukuk instead rep-

two important differences. Firstly, the venture must be made in a halal ac-

resent a profit and risk sharing agreement between the issuer and the

tivity or underlying asset. This means that for example, a companys core

investor whereby sukuk holders own a proprietary (proportional) inter-

business must relate to a halal sector such as computers, energy, textiles,

est in the sukuk underlying assets. Shariah compliance is attained by

or transportation. If a companys core business relates to haram activities

ensuring that financial returns are linked to actual revenues generated

such as alcohol (e.g., bars) or gambling (e.g., casinos) then this would be

by the performance of the underlying sukuk assets, instead of fixed in-

unacceptable under Islamic law. Nevertheless, in practice peripheral ha-

come returns. Therefore, while sukuk may often be referred to as Islamic

ram activities have been accepted by SSBs in certain jurisdictions, mean-

finance bonds, in practice their underlying structure is very different

ing that for example, so long as the haram activity accounts for no more

to traditional Western fixed income bonds. Sukuk may be structured as

than 5% (10% for some SSBs) of a companys revenue or gross income,

either asset-based (transfer of beneficial ownership) or asset-backed

then that will not constitute a core business and will be allowed under

(transfer of actual ownership or true sale), but all sukuk usually involve

Shariah rules. However, this rule has not been applied consistently across

the use of a (bankruptcy-remote) special purpose vehicle to originate or

all Islamic finance jurisdictions and SSBs, meaning that in practice the ex-

issue sukuk or trust certificates. Asset-backed sukuk therefore present

act requirements (e.g., in terms of divestiture of haram investments) may

Islamic investors with additional ownership risk, although asset-based

differ according to jurisdictional and SSB dictates. Secondly, it should be

sukuk are generally more risky to Islamic investors in terms of the oc-

noted that the final distribution of profits under a Mudaraba can only take

currence of default or bankruptcy. Consequently, while traditional West-

place after the final winding down of the Mudaraba investment.

ern bonds are essentially contractual debt obligations, sukuk holders


are instead entitled to share in the profits generated from the underlying

Musharaka (partnership contract)

assets (as well as share sale proceeds of the assets upon liquidation),

The Musharaka (or Musharakah) contract functions in much the same way

meaning there is no guaranteed return on principal for sukuk [Akkizidis

as the Mudaraba contract as it involves a joint venture between two or

and Khandelwal (2008)]. This different structuring means that whereas

more parties acting together as partners. However the Musharaka is dis-

Western bond holders do not share on losses relating to bonds, Islam-

tinguished from the Mudaraba in that all parties may contribute financial

ic sukuk holders are exposed to additional financial risk as they must

investments in the joint venture, and they may all take part in its manage-

share in any losses which occur on the underlying ownership of the

ment. Nevertheless, in order to be Shariah compliant profits may be allo-

assets. This impacts upon the risk profile of Islamic sukuk instruments,

cated according to pre-determined ratios, but may not be pre-fixed in ad-

meaning that while AAA rated Western long-term bonds are seen as low

vance as this would preclude risk taking. Furthermore, any losses made in

risk long-term investments, the risk profile of economically equivalent

the venture may be determined in accordance with the proportion of their

Islamic sukuk will be different because of the general lack of second-

initial financial investments. Musharaka contracts have become especially

ary markets for sukuk issues, together with the increased financial risk

prominent in retail finance in the form of diminishing Musharakas, where

exposure of the underlying assets (e.g., as witnessed in the defaulted

they are regularly used to fund the purchase of property through home pur-

sukuk payment of the UAEs Nakheel sukuk in 2008-2009 owing to the

chase plans. Essentially, these involve a property being divided into units

Dubai property crash). In some instances recourse to underlying assets

and one party to the contract gradually buying all the units from the other

or cash flows has proven difficult, and consequently recent efforts have

party over a fixed period (e.g., 15 years), at which point full ownership of

been made to use bankruptcy remote structures to ensure asset priority

the property will be transferred. However, Akkizidis and Khandelwal (2008,

of sukuk certificate holders vis--vis creditors when sukuk default [Khni-

14) observe that [u]nequal distribution of wealth is treated as a sin in Islam

fer (2010a), (2010b)]. Finally, although sukuk can be listed on recognized

and hence any partnership with unjust distribution of profits or losses is

exchanges, it has been observed that [u]nlike in the conventional bond

not permitted. Consequently they state that the Musharaka contract is

market, however, sukuk tend to be held to maturity and the secondary

seldom used because of the high degree of uncertainty over the partner-

market is not very active. Although quotes are provided by some market

ship returns, and thus it is predominantly used in cases which involve large

makers, the spreads between bid and asking price are particularly wide

investments and for larger joint-venture projects [Akkizidis and Khandelwal

and availability of issue is still thin [Schoon (2008), 634].

(2008)]. The Musharaka partnership can be terminated upon notice, or al20

ternatively by the death or incapacity of a partner.

The Capco Institute Journal of Financial Transformation


Enhancing Islamic Finance through Risk Benchmarking

Ijara (lease contract)

institutions major risks. This is because they argue that peoples tenden-

Ijara is a Shariah compliant leasing contract which is widely used in Islamic

cy to compromise can lead to an excessively long list of risks, therefore

finance for both operational lease (ijara), or finance lease (with purchase

the onus is on the institution to identify the few risks that really matter,

option) (ijara wa iqtina) purposes. In order to be Shariah compliant the

which is typically four to six key risks most commonly encountered (i.e.,

asset subject to lease must continue to hold value throughout the dura-

demand, commodity, country, operational, and foreign exchange (forex))

tion of the lease, it must be sufficiently certain (i.e., no consumable goods

[Buehler et al. (2008)]. This step is followed by deciding whether or not a

such as oil); and the lessor must hold the risk of ownership (i.e., liability for

company is the natural owner of the risk involved, since natural owner-

maintenance and insurance) throughout the term of the lease. Also, if the

ship assessment yields a clear corporate risk strategy which can create

leased asset is destroyed or the lessor is prevented from using the leased

superior returns [Buehler et al. (2008)]. Moreover, other non-natural identi-

asset then the ijara contract is terminated and the parties are released from

fied risks can be more cost-effectively hedged or transferred to third par-

their obligations under the lease. Chinoy (1995) states that in order to con-

ties owing to their superior scale and knowledge economies [Buehler et

form to Shariah the ijara is based on profit sharing, so that rentals consti-

al. (2008)]. In fact, Gerken et al. (2010) believe that getting risk ownership

tuting the Islamic banks return are not calculated on the basis of capital

right in the post-financial crisis new risk paradigm is crucial, since in

involved, but instead on the profitability of the asset once it is employed

recent years many firms owned risks that they either had not considered

by the lessee, i.e., rental payments are a fixed percentage of annual profits

or for which they were ill suited. It is therefore argued that what is needed

attributable to leased asset output. Chinoy (1995) further adds that Islamic

is that an institution takes on board only those risks for which it has three

ijara leasing has the potential for Shariah securitization (under a Shariah

core lines of defense, namely (i) a resilient and flexible business model;

certificate of debt) because the funding can be totally matched from a li-

(ii) well-developed risk skills and capabilities; and (iii) sufficient financial

quidity perspective to the asset being purchase, and the cash flow is rela-

strength to absorb risk if it should materialize [Gerken et al. (2010)]. In

tively certain. In practice ijara leases are commonly used as part of Islamic

consideration of all of these points, the discussion here will concentrate

mortgage transactions as buyers lease the house throughout the duration

on the three major risk groupings faced by all major banks and credit

of the mortgage, make lease and capital repayments throughout, and title

institutions, namely market risk, credit risk, and operational risk [Ioan-

to the property is transferred at the end of the mortgage term.

nis et al. (2006)]. Additionally, Shariah compliance risk, which is specific


to Islamic finance institutions and products, will also be discussed and

An overview of financial risk and financial risk


modeling

compared to reputational and compliance risks faced by Western banks.

Within the fields of economics and finance there exist a plethora of both

Market risk

well known and less familiar risk typologies. These include, but are not

Market risk is defined by the Bank for International Settlements (BIS)

limited to, acquisition risk; commodity risk; country risk; credit risk; di-

as the risk of losses in on and off-balance-sheet positions arising from

saster recovery risk; environment risk; equity market risk; inflation risk;

movements in market prices, with these risks being comprised of risks

interest rate risk; investment risk; legal risk; liquidity risk; operational risk;

relating to interest rate related instruments and equities in the trading

political risk; regulatory risk; reputational risk; settlement risk; systemic

book, and forex and commodities risks throughout the bank [BIS (2006),

risk; and volatility risk. In practice, certain types of risk can be signifi-

157]. Market risk may be comprised of sub-risks which make up the total

cantly mitigated or reduced, for instance diversifiable risk can refer to

market risk faced by a bank or financial institution. It will also reflect the

certain security investment risks that can be virtually eliminated through

specific nature of the business undertaken (e.g., investment or market

the diversification of a securities portfolio. On the other hand, other types

trading), and the nature of the business undertaken will in turn affect the

of risk such as systemic risk, which may affect the interlinkages and

weighting of any sub-risks faced. For instance, sub-risks may include cur-

interdependencies of an entire financial system, are of a much greater

rency risk (risk of variation of forex rates); commodity risk (risk of variation

magnitude, and therefore cannot generally be significantly mitigated at a

of commodity prices); equity risk (risk of variation of stock or stock index

micro firm level. Therefore, in order to develop an accurate Islamic bank-

prices); interest rate risk (risk of variation of interest rates); and liquid-

ing financial risk benchmarking system, it is first necessary to set the

ity risk (risk that the price at which you buy (or sell) something may be

groundwork by developing a financial risk model that best reflects the

significantly less advantageous than the price you could have achieved

actual financial risk faced by financial institutions and financial products.

under more ideal conditions [Allen (2013), 3]). Thus, where financial

This can be done using the same methodology as would be required for

instruments are not exposed to stock price movements or indexes, or

devising and implementing a banking risk management system.

do not involve commodities, there will be no equity or commodity risks


weightings in the overall market risk mix. Within economics and finance

In this regard Buehler et al (2008) argue that the first step in developing

it is stated that credit and financial institutions now widely use value-

an effective risk management system is identifying and understanding an

at-risk (VaR) models in a bid to measure losses on a constant portfolio

21

over a specified period (often the next day or next 10 days) that will be

usually only available for larger commercial enterprises, companies, or fi-

exceeded only on a given fraction of occasions (typically a probability

nancial institutions. Therefore, in practice BFIs have generally developed

level of 1%) for internal risk management purposes (i.e., maximum loss

internal methodologies for assessing credit risk. Alternative approaches

likely to occur over a pre-specified time period for a given confidence

have been developed and used which rely on more objective methodolo-

level) [Ioannis et al. (2006), 1-2]. VaR is said to be precisely derived from a

gies to modeling credit risk, including the structural approach, the re-

distribution of changes with probability () in the market value of portfolio

duced form approach, and the actuarial based approach. The struc-

(P) over time horizon (T) (Figure 1) [Medova and Kyriacou (2001)].

tural approach (e.g., CreditMetrics, KMV) seeks to model credit


risk through the correlation of credit default risk and capital structure,

There are three further points to be noted about VaR. The first is that it

although it is typically subject to a number of limitations (e.g., assump-

is derived from the Capital Asset Pricing Model (CAPM), which is itself

tion of no market risk over defined period or non-economy dependent

based on a number of simplified assumptions (i.e., many investors behav-

default probabilities) [Black and Scholes (1973), Merton (1974, 1977)].

ing competitively and all using risk-free rate, single period time horizon,

It is also argued that this approach estimates the correlation between

no taxes or commissions). The second is that many of the techniques

asset-based rates of return using equity returns (as asset returns are un-

for measuring VaR are based on unrealistic or non-empirically supported

observable), meaning that this VaR methodology is sensitive to correla-

assumptions (e.g., distributional assumption; stationarity; non-negativity

tion co-efficients, yet the accuracy of the methodology cannot be verified

requirement; random walk assumption of intertemporal unpredictabil-

[Jarrow and Turnball (2000)]. The reduced form model of credit default

ity [Allen et al. (2003)]). The third point is there are different quantitative

approach is considered to assume that a firms default time is inacces-

methods for calculating VaR and the method used is dependent on the

sible or unpredictable and driven by a default intensity that is a function

portfolio type and the data available for modeling, e.g., variance-co-

of latent state variables [Arora et al. (2005)]. Default probabilities are nor-

variance (VcV) (or delta-normal) and multivariate Monte Carlo simula-

mally estimated using explanatory variables (e.g., counterparty-specific

tion methods are based on parametric estimation, while the historical

and non-counterparty-specific macroeconomic variables) together with

simulation method is based on non-parametric estimation.

historical default data [Jarrow and Turnball (1995), Hull and White (2000)].
Arora et al. (2005) believe that while this type of model does not com-

Credit risk

promise on the theoretical issue of complete information, it is limited in

Credit risk or counterparty risk is the risk that a counterparty to a

that there is a lack of clear economic rationale for defining the nature of

credit transaction will fail to meet its payment obligations under the terms

the default process. The actuarial based approach (e.g., CreditRisk+)

specified in a credit agreement, thereby resulting in a delay or loss to the

models credit risk as default loss using expected default rate risk drivers,

bank or financial institution. This may in turn affect a financial institu-

and using a function of shared risk factors such as default rates and vola-

tions liquidity, which represents a real and significant risk to banks, since

tilities. The modeling methodology used by a credit institution will reflect

more than 70% of a banks balance sheet relates to credit risk manage-

relevant factors such as the portfolio type, obligor profile, credit policies,

ment [van Greuning and Iqbal (2008)]. A widely used source of measuring

and business cycle [Ioannis et al. (2006)].

credit risk is by utilizing credit ratings (i.e., forward-looking opinions of


credit risk) available from an established provider such as Fitch Ratings,

Operational risk

Moodys, or Standard & Poors. However, these types of credit ratings are

Operational risk is defined by the BIS as the risk of loss resulting from
inadequate or failed internal processes, people and systems or from external events, and while legal risk3 is included within the definition, stra-

Pr[P(T) -VaR]=
F-1 () = VaR (m=0)

tegic and reputational risk are excluded [BIS (2006), 144]. Operational risk
may include sub-risks such as accounting risk; enterprise risk; funding
liquidity risk; legal risk; operations risk; and technology and systems risk.
The chief difficulty with identifying operational risk is its extensive scope
and lack of homogeneity among different BFIs. For instance, people risks
may be dependent on the size of the firm, and technology risks may
be dependent on the firms technology investment and infrastructure. Indeed Akkizidis and Khandelwal (2008) have noted that the wide range

VaR

22

Figure 1 Value at Risk distribution curve [Medova and Kyriacou (2001)]

3 The BIS states: Legal risk includes, but is not limited to, exposure to fines, penalties, or
punitive damages resulting from supervisory actions, as well as private settlements [BIS
(2006), 144].

The Capco Institute Journal of Financial Transformation


Enhancing Islamic Finance through Risk Benchmarking

of activities included within operational risk make it difficult to measure

law. In consequence, the contract could be declared (partially) void in a

and manage, and also to apply a standard organizational model. Never-

Shariah court. In practice Shariah compliance is said to be what lends

theless, BFIs have been developing models which rely on a similar set

financial products or services their legitimacy within Islamic markets, as

of factors, including internal audit ratings and control self-assessments,

Muslim consumers, investors, and clients are concerned with Shariah

as well as operational risk indicators (i.e., income volatility, loss experi-

compliance as Islam restricts consumption to that which is good and

ence, rate of errors, turnover, volume) [Akkizidis and Khandelwal (2008)].

wholesome (hala/ un tayyib), but prohibits that which is foul, harmful,

From one perspective it has been argued that operational risk is usually

unjust, or sinful [DeLorenzo (2007)]. This is unique to Islamic finance and

defined in the negative (as it includes all risks which are not categorized

differentiates it from its Western finance counterparts. For instance, while

as either market or credit risks), but that in any event it comprises virtu-

the fundamental objective of a Western mortgage product may be to pro-

ally all the risk that cannot be managed through the use of liquid mar-

vide a long-term financing solution through the payment of interest, a

kets, meaning it is excluded from the scope of financial risk management

comparative Islamic mortgage product would, in addition to the provi-

[Allen (2013)]. Nevertheless, the BIS has developed an operational risk

sion of long-term financing, necessarily have to comply with the funda-

framework which is applicable to the measurement of minimal capital

mental objective of full compliance with all pertinent Shariah principles,

standards, with three methods employed to calculate operational risk

rules, and obligations. Otherwise a Muslim consumer would risk offend-

capital charges in a continuum of increasing risk sensitivity and sophisti-

ing against Islam, and would in principle breach his or her trust with Al-

cation [BIS (2006)]. These methods (the Basic Indicator Approach, the

lah.4 Consequently Shariah compliance through the issuing of Islamic

Standardised Approach, and the Advanced Measurement Approach)

fatwas (Islamic religious rulings issued by qualified Muslim scholars) and

equate capital charges with either annual gross income (previous three

SSB rulings, therefore provides acceptance and validity from a Shariah

years); use regulatory capital charges as a proxy for operational risk

perspective, a Shariah religious Kitemark if you will. Nevertheless, SSB

exposure across eight business lines; or adopt internal qualitative and

rulings may only be applicable in a particular jurisdiction, or may only

quantitative operational risk models [BIS (2006)]. The BIS approach how-

apply to Islamic institutions that fall under the purview of any particular

ever, does not include reputational risk which is a significant risk in the Is-

SSB. Consequently, what may be perceived as Shariah compliance by

lamic banking model. There are, however, a number of other operational

one SSB may not be accepted by other SSBs acting in different jurisdic-

risk modeling approaches which have also been developed (Table 1: Ap-

tions, or comprised of different Muslim schools of thought. This lack of

proaches to operational risk modeling [Smithson and Song (2004), 58]).

global uniformity is a particular difficulty that is faced by Islamic finance


organizations and institutions around the world, however, many interna-

Compliance or Shariah risk

tional Islamic institutions such as the Islamic Financial Services Board

The issue of Shariah risk (or Shariah compliance or non-compliance risk)

(IFSB) and The Accounting and Auditing Organization for Islamic Finan-

is unique to Islamic finance institutions and instruments, since Shariah

cial Institutions (AAOIFI) have been making significant efforts to introduce

compliance is sine qua non of Islamic finance. In short Shariah risk can

global common Shariah standards. Rethel (2011, 95) therefore argues

be simply defined as the risk that arises from a failure to comply with

that Shariah compliance does not automatically entail the ethical and

Shariah rules and principles, and applies just as much to Islamic BFIs as

social objectives of Islamic finance, namely risk-sharing and participatory

it does to Islamic finance products. Alternatively, Sol (2007, 4) defines

economic growth, so that input and output legitimacy are often not fully

it as the risk that the terms in the agreed in a contract do not effectively

congruent. Nevertheless, Shariah risk still carries with it not only person-

comply with Islamic jurisprudence and thus are not valid under Islamic

al repercussions for the Muslim customer, but also reputational repercussions for the offending Islamic bank or financial institution, since breach
of Shariah rules may cause significant reputational harm and censure

Techniques for quantifying operational risk

by the relevant Muslim community. Ayub (2008) therefore observes that

Process approaches

Factor approaches

Actuarial approaches

SSBs play a crucial role in ensuring Shariah compliance within particu-

Causal

Risk indicators

Empirical loss distributions

lar jurisdictions and for Islamic institutions operating under the guidance

CAPM-like models

Explicit distributions parameterised using historical


data

Predictive models

Extreme value theory

Bayesian belief networks


Fuzzy logic

of any particular SSB. For instance, this may translate in practice for

Statistical quality control


and reliability analysis
Connectivity
System dynamics

Table 1 Approaches to operational risk modeling [Smithson and Song


(2004), 58]

an exemplary SSB to include ensuring Shariah compliance criteria for

4 An-Nahl (16 Verse 27): Then, on the Day of Resurrection, He will disgrace them and will
say: Where are My (so called) partners concerning whom you used to disagree and
dispute (with the believers, by defying and disobeying Allah)? Those who have been given
the knowledge (about the Torment of Allah for the disbelievers) will say: Verily! Disgrace
this Day and misery are upon the disbelievers.

23

prohibited services (e.g., investment in restaurants, casinos and hotels


with bars for prohibited activities), as well as monitoring investments in
companies (i.e., ensuring interest income is no more than 5%, debt ratio

Classification

Type

Meaning

Nature of
relevant other

Internal

Comparing within one organization about the


performance of similar business units or processes

Competitor

Comparing with direct competitors, catch up or even


surpass their overall performance

Industry

Comparing with company in the same industry,


including non-competitors

Generic

Comparing with an organization which extends


beyond industry boundaries

Global

Comparing with an organization where its


geographical location extends beyond country
boundaries

Process

Pertaining to discrete work processes and operating


systems

Functional

Application of the process benchmarking that

(leverage) is no more than 10-33%, and total illiquid assets are less than
10-33% of its total assets).

An overview of benchmarks and financial risk


benchmarking
Benchmarking has certainly come a long way since its early application within specialist manufacturing fields [Tucker et al. (1987), McNair and
Leibfried (1992)]. Indeed, nowadays benchmarking spans the whole gamut
of industry sectors, including areas such as aerospace, construction, de-

Content of
benchmarking

fense, education, finance, health, shipping, technology, transportation, and


utilities, to name but a few. But what exactly does it entail? In the words of

compares particular business functions in two or


more organizations

Camp (1989): Benchmarking is the search for the best industry practices
which will lead to exceptional performance through the implementation of
these best practices. Delpachitra and Beal (2002) opine that while modern
benchmarking involves comparing the performance of organizations, the
primary objective of benchmarking is to increase the probability of attaining competitive advantage. On the other hand, Anand and Kodali (2008)
define benchmarking as a continuous analysis of strategies, functions,
processes, products or services, performances, etc. compared within or
between best-in-class organizations by obtaining information through ap-

Purpose for the


relationship

Performance

Concerning outcome characteristics, quantifiable in


terms of price, speed, reliability, etc.

Strategic

Involving assessment of strategic rather than


operational matters

Competitive

Comparison for gaining superiority over others

Collaborative

Comparisong for developing a learning atmosphere


and sharing of knowledge

Table 2 Classification of benchmarking [Fong et al. (1998), 410]

propriate data collection method, with the intention of assessing an organisations current standards and thereby carrying out self-improvement by

In light of this Moriarty (2011, 598) argues that [a] benchmarking process

implementing changes to scale or exceed those standards. Consequent-

must identify organizational competencies, gauge their value or impact

ly, while the underlying thrust of these definitions allude to performance

according to some consistent metric (cardinal, real, monetary, etc.) and

measurement, comparison, and improvement, in practice much depends

also establish how these competencies contribute to the sustainability of

on the particular type (internal or external) and area to which benchmarking

the exemplar organization. In practice benchmarking may take a variety

relates. For example, modern practices may relate to a plethora of bench-

of forms, but for present purposes it is helpful to briefly examine an illus-

marking types, including inter alia collaborative, comparative, financial,

trative benchmarking model within the banking sector.

functional, generic, operational, performance, process, product, and strategic benchmarking typologies (Table 2).

Data Envelopment Analysis (DEA) is defined as a linear programming


technique which is used for measuring relative efficiency for a set of ho-

24

More specifically within the field of finance, financial benchmarking

mogenous decision-making units, in converting multiple inputs (resourc-

might, for example, involve the use of financial analysis to compare and

es) to produce multiple outputs (performances) [Mukherjee et al. (2002)].

contrast a firms overall operating conditions with a view to compara-

It is noted that DEA has been used extensively as a way of studying

tively evaluating its competitiveness, efficiency, productivity, and quality

the performance of the banking sector (i.e., performance efficiencies) in

levels. In order to do this a firm might employ key performance indicators

the last decade [Mukherjee et al. (2002)]. Mukherjee et al. (2002) under-

(KPIs) or other metrics such as gross, operating, and net profit margins;

took a study of performance benchmarking and strategic homogeneity

price/performance ratios; price/customer satisfaction portfolio; revenue

of Indian banks by using DEA to develop a method of benchmarking the

to fixed assets ratios; revenue or cost per employee; or sales and profit-

performance of Indian banks using their published financial information.

ability trends. From the outset it should be noted that what is important

Performance was defined as how a bank was able to utilize its resources

is that key financial benchmarking attributes are expressly identified and

to generate business transactions, which was measured by the ratio and

delineated in a benchmarking operational plan. At a minimum this will in-

labeled as each individual banks efficiency [Mukherjee et al. (2002)]. The

clude, for example, establishing benchmarking objectives; longevity; op-

benchmarking study used five input parameters (net worth, borrowings,

erating parameters; comparator data, factors or metrics; comparator or

operating expenses, number of employees, number of branches) and five

best-in-class firms or sectors; baseline levels; and benchmarking tools.

output variables (deposits, net profits, advances, non-interest income

The Capco Institute Journal of Financial Transformation


Enhancing Islamic Finance through Risk Benchmarking

(e.g., sale of investments), interest spread) [Mukherjee et al. (2002)]. What


1. Identify benchmarking subject

was interesting about the study was that the DEA clustering highlighted
poor-performing clusters, enabling the identification of weak performing
banks, which were subsequently probed and found to have underlying

Planning

problems relating to overstaffing, high percentages of non-performing

2. Identify benchmarking partners

3. Determine data collection method and collect data

assets, high costs of funds, and low capital adequacy ratios [Mukherjee
et al. (2002)].

Analysis

This highlights the point that the importance of benchmarking does not

4. Determine current competitive gap

5. Project future performance

simply lie in the process itself, but rather in the final interpretation of the
6. Communicate findings and gain acceptance

benchmarking results obtained. Indeed, Wendel (1993) believed that the


value of benchmarking often results from the questions raised in the

Integration
7. Establish functional goals

process, for example, why is there variance between similar groups or


individuals, or do differences result from factors outside the control of

8. Develop action plans

management? However, at the same time it has also been argued that
criticisms of benchmarking include that it is naturally an a posteriori or-

Action

9. Implement plans and monitor progress

ganizational practice and furthermore, that slavish adoption of exemplary


practices engenders uncompetitive homogeneity [Moriarty (2011)]. It can

10. Recalibrate the benchmark

also be argued that there are inherent difficulties peculiar to benchmarking, including the accuracy and availability of comparator data or factors,
since the most apposite benchmarking may sometimes require obtain-

Figure 2 Xerox benchmarking model [Camp (1989); Anand and Kodali


(2008), 270]

ing valuable competitor market data. Yet this contradicts one of the fundamental underlying objectives behind benchmarking, namely attaining

effort, external advisors, and costs); (ii) processes (i.e., risk identifica-

competitive advantage over market competitors. Indeed, Standard and

tion, assessment and management implementation, audits, risk transfer,

Poors Risk-to-Price : The New Risk Benchmark highlights exactly

financing, and modeling); (iii) outputs (i.e., number of risk assessments);

why valuable benchmarking may often involve proprietary information

(iv) feed-back (i.e., evaluating the effectiveness and awareness of mea-

[Standard and Poors (2013)]. This risk benchmark is said to be a relative

sures); (v) feed-forward (i.e., targets and objectives, risk-based planning,

measure of how well a security may be compensating its owner, through

event horizon scanning); (vi) monitoring (i.e., assessing payback); and (vii)

yield, for the embedded market and credit risks [Standard and Poors

governance (i.e., risk strategy setting process, organizational incentives,

(2013)]. In actuality what this involves is the daily calculation of propri-

policy inhibitors). This is similar to the ten-step benchmarking design

etary Risk-to-Price (R2P) scores for corporate debt issues through the

framework initially proposed by Camp (1989), which incorporated four

evaluation of their probability of default, volatility, and option-adjusted

procedural stages, including planning, analysis, integration, and action

spread [Standard and Poors (2013)]. This R2P risk benchmarking typol-

(Figure 2).

ogy therefore allows the identification of potential pricing errors between


overall bond pricing spreads and the bonds risk components [Standard

Although there exist a wide range of benchmarking types, within finan-

and Poors (2013)].

cial risk benchmarking there are certainly recognized procedures, frameworks, and inputs that can be utilized. The difficulty in practice however,

According to Mainelli (1999), as a business process, risk management it-

is that these types of benchmarks are normally highly subjective owing

self is well suited to benchmarking, although it does have certain idiosyn-

to their intended use, namely the attainment of competitive advantage.

crasies. For instance, since risk management is a relatively new process,

Thus Harry Lipman, the Global Product Manager for Bloomberg OTC

the setting of standards (benchsetting) may be as much a part of the

Derivatives, has argued that given that financial benchmarks determine

process as comparisons [Mainelli (1999)]. Furthermore, risk management

payments on, and set prices for, at least US$750 trillion worth of finan-

must be viewed as a holistic process affecting the entire organization, and

cial products (i.e., mortgages, loans, credit derivatives), achieving the

since risk metrics are still relatively in their infancy, much work may need

appropriate balance to regulate financial benchmarks is crucial [Lipman

to be done in terms of delivering quantitative results, as opposed to qual-

(2013)]. Lipman (2013) proposed that financial benchmarks be devel-

itative comparisons [Mainelli (1999)]. Consequently, Mainelli (1999) posits

oped with a strong, consistent, and transparent design that creates a

a seven element approach to risk benchmarking, including consideration

level playing field for all, and furthermore that they should be acces-

of (i) inputs (i.e., expenditure on risk management, compliance time and

sible on fair, non-discriminatory and reasonable terms. It was argued that

25

this could be achieved through the implementation of benchmarks which


address three key issues, namely (i) the creation of non-discriminatory
access; (ii) the avoidance of monopolies via exclusive licenses; and (iii)
ensuring that constituent and weighting information is available to all
key participants [Lipman (2013)]. This line of argument therefore alludes
to financial benchmarks being developed in a cohesive and transparent manner, which is more akin to collaborative style benchmarking as
opposed to solely competitive style benchmarking. Within the current
context of Islamic finance it is proposed that this type of collaborative
style benchmarking is to be preferred, especially since it would uphold
the underlying ethos of Shariah by ensuring an ethical and fair approach
to financial and risk benchmarking within Islamic finance initiatives.
Taking into account the crux of these concerns, the International Organi-

What specific areas of risk function responsibility


are covered in the risk reporting package?
Market risk
Operational risk
Liquidity risk
Credit portfolio monitoring
ALM risk
Capital adequacy
Limits monitoring
ICAAP process
Economic capital
Credit analysis
Compliance/legal risk
Credit approvals

95%
93%
90%
90%
79%
72%
65%
59%
59%
50%
43%
41%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90%100%

Figure 4 Scope of risk reporting [PricewaterhouseCoopers (2011), 4]

zation of Securities Commissions (IOSCO) proposed new Principles for


Financial Benchmarks (the IOSCO Principles) via a Consultation Report
in order to create an overarching framework of principles for benchmarks
used in financial markets [IOSCO (2013)] (Figure 3). It is therefore these
all-encompassing principles that will be used to provide the general
framework for the financial risk benchmarking of Islamic finance products. Firstly, the IOSCO Principles define a benchmark as incorporating
prices, estimates, rates, indices, or values that are (i) made available to
users (whether free of charge or for payment); (ii) calculated periodically
(entirely or partially by the application of a formula or another method
of calculation) for the value of one or more underlying interests5; and (iii)
used for reference purposes. Reference purposes include one or more
of: (i) determining the interest payable, or other sums due, under loan
agreements or under one or more financial contracts or instruments; (ii)

How would you assess the quality and relevance of


data used in your bank's risk management?
High-quality data flow and data
governance (including control processes)
High-quality data flow but
limited data governance

24%

Average quality data flow and


data governance

41%

Some issues related to quality of


data and governance
Significant effort is required to
achieve appropriate quality of data

0%

Series 1

17%

13%
5%

10% 20% 30% 40% 50%

Figure 5 Data quality and governance [PricewaterhouseCoopers (2011), 7]

determining the price at which a financial instrument may be bought or


sold or traded or redeemed (or the value of a financial instrument); or (iii)
measuring the performance of a financial instrument [IOSCO (2013)]. The

Governance

26

Principle 1 - overall responsibility of the administrator


Principle 2 - oversight of third parties
Principle 3 - conflicts of interest for administrators
Principle 4 - control framework for administrators
Principle 5 - internal oversight

Quality of the
benchmark

Principle 6 - benchmark design


Principle 7 - data sufficiency
Principle 8 - hierarchy of data inputs
Principle 9 - periodic review

Quality of the
methodology

Principle 10 - content of the methodology


Principle 11 - changes to the methodology
Principle 12 - transition
Principle 13 - submitter code of conduct
Principle 14 - internal controls over data collection

Accountability

Principle 15 - complaints procedures


Principle 16 - audits
Principle 17 - audit trail
Principle 18 - cooperation with regulatory authorities

Figure 3 Principles for financial benchmarks [IOSCO (2013)]

IOSCO principles contain 18 principles covering four areas, including: (i)


governance; (ii) quality of the benchmark; (iii) quality of the methodology;
and (iv) accountability. The governance principles seek to put in place
appropriate governance arrangements to protect the integrity of the
benchmark and to address conflicts of interest [IOSCO (2013)]. The quality principles aim to promote the quality and integrity of benchmark determinations in order to reflect a credible market for the relevant interest
[IOSCO (2013)]. The methodology principles seek to promote the quality
and integrity of the methodologies6 by providing minimum requirements

5 Interest is defined: Refers to any physical commodity, currency or other tangible goods,
intangibles (such as an equity security, bond, futures contract, swap or option, interest
rates, another index, including indexes that track the performance of a rule-based trading
strategy or the volatility of a financial instrument or another index), any financial instrument
on an Interest, which is intended to be measured by a Benchmark. Depending on the
context, it is assumed that the word Interest also includes the market for such Interest.
6 Methodology is defined: The written rules and procedures according to which
information is collected and the Benchmark is determined.

The Capco Institute Journal of Financial Transformation


Enhancing Islamic Finance through Risk Benchmarking

for published information regarding the benchmark [IOSCO (2013)]. The


accountability principles seek to apply appropriate complaints process-

Islamic
BFI (1)

es, documentation requirements and audit reviews to provide sufficient


evidence of benchmark compliance [IOSCO (2013)].
It can also be argued that there exists a strong need for the development

Islamic
BFI (6)

of more widely recognized financial and risk benchmarking standards.

RBRs

For instance, PricewaterhouseCoopers (2011) undertook a risk manage-

ISI

ment benchmarking survey in order to benchmark the performance of


risk management functions at banks, by comparing risk management
trends and capabilities in 78 banks across 20 countries from Europe, the

Islamic
BFI (2)

RBRs
RBRs

RBRs

RBRs

Islamic
BFI (5)

Islamic
BFI (3)

RBRs

Middle East, Africa, and India. The survey found that risk management
in most institutions still tended to operate in isolation rather than being

Islamic
BFI (4)

integral to the broader business processes, operations, and behavior of


the organization [PricewaterhouseCoopers (2011)]. Furthermore, it was
found that many banks were failing to capture newer areas of focus (e.g.,
credit process risk, compliance risk, legal risk) (Figure 4). It was argued

Figure 6 Internal Level IFRB framework

that the siloed view of risk was often exacerbated by the lack of integration between risk and finance functions (60% of participants disclosed
Islamic
BFI (1)

limited risk-finance interaction), and that [t]his impedes the banks abilW
Western
BFI (1)

ity to develop a risk-adjusted view of its performance, understand the

Islamic
BFI (2)

capital implications and take account of the potential impact of market


stresses and other future risks [PricewaterhouseCoopers (2011), 4].

RBRs
W
Western
BFI (3)

W
Western
BFI (4)

RBRs

RBRs

RBRs

ing of risk within Islamic finance is not an individual endeavor, but is


risk benchmarking loses its significance, credibility, and influence, and

ISI

Islamic
BFI (4)

RBRs

Applying risk benchmarking to Islamic finance

rather envisaged as a collective undertaking. Otherwise Islamic finance

Islamic
BFI (3)

RBRs
RBRs

of participant banks believed that the quality and governance of their

It should be emphasized from the outset that the theoretical benchmark-

RBRs

RBRs

Data quality was also found to be a critical concern, as only around 40%
data which informed their risk models was high (Figure 5).

RBRs

W
Western
BFI (2)

RBRs
W
Western
BFI (5)

Islamic
BFI (5)

RBRs

W
Western
BFI (6)

Islamic
BFI (6)

BFIs may end up rolling out unrealistic risk models and risk management practices, solely to achieve low risk ratings. Consequently it is to be
undertaken and adopted in the spirit of Shariah in order to advance an

Figure 7 External Level IFRB framework

ethical and fair approach to the future regional and global development
of Islamic finance. In particular it would seek to advance the secondary

significance. Therefore for the Internal Level, the IFRB framework would

objectives of Shariah, including the establishment of justice and equity

be implemented across each of six Islamic BFIs for a set period, thereby

in society; the promotion of social security, mutual help and solidarity;

resulting in the development of Risk Benchmark Ratings (RBRs) for each

and the promotion of moral values and cooperation in matters of good-

Islamic BFI. The RBRs would be collected and monitored by an Inde-

ness [Ayub (2008)]. From a high level perspective, the Islamic Finance

pendent Supervisory Institution (ISI) that would have legal and opera-

Risk Benchmark (IFRB) framework design is relatively simple, yet it of-

tional safeguards in place to protect all confidential financial and other

fers significant benefits. Firstly, although the IFRB framework is designed

types of information supplied. The ISI itself would be funded by financial

to operate optimally across a mixture of both Islamic and Western BFIs,

subscriptions made by all participating BFIs, but it would still retain op-

it can first operate at an interim level if need be. Both the first interim

erational independence in much the same way that supranational institu-

Internal Level (Figure 6) and the final External Level (Figure 7) are

tions such as the International Monetary Fund operate (e.g., payment of

intended to each operate with an envisaged minimum of six BFIs. This

financial quotas by its members). The RBRs would allow all the Islamic

would allow risk benchmarking comparisons to have sufficient relative

BFIs to compare their relative institutional risk weightings, as well as

27

relative risk weightings across specific asset classes, and let them see if

Enhancing the legitimacy of Islamic finance

they were operating at excessive risk levels as compared to other Islamic

Ahmad (1993) argues that Muslim economists have put forth theoretical

BFIs. From one perspective the argument can be made that this Internal

models of Islamic banking based on the principle of Mudarabah, to justify

Level framework might only distort true comparisons and benchmarking

claims of superiority over interest-based commercial banking products,

efforts vis--vis Western BFI counterparts. Nevertheless, it is contended

especially in regards to equity, efficiency, stability, and growth. Indeed it is

here that the Internal Level framework would only be intended to oper-

recognized that a common perception about Islamic banking is that Islam-

ate as an interim or provisional measure in order to first allow Islamic

ic BFIs carry less risk owing to the fact that they are not based on interest

BFIs to better develop their collaborative risk management capacities

rates (i.e., less volatility), and that most Islamic finance products are trade

and expertise, and in order to allow themselves time to jointly improve

financing instruments based on mark-up arrangements [Toumi and Viviani

their risk measurement and management systems. It is also intended as

(2010)]. However, at the same time it has also been argued that Islamic

a facilitative step, as in reality it might be unlikely that Islamic BFIs would

banks are subject to the same risks (credit, market, operational) as West-

be willing to participate in a risk benchmarking scheme with Western

ern banks, except that they are magnified because Islamic products are

counterpart BFIs, without first assessing their own risk management ca-

new [Toumi and Viviani (2010)]. Furthermore, Islamic BFIs face unique risks

pabilities and risk levels. In this way the Internal Level framework would

owing to the specific assets and liabilities structure of their balance sheets,

act as a type of primer platform, from which Islamic BFIs could move on

including: (i) fiduciary risk (risk of misconduct, negligence, or contrac-

to realistically compare relative risk weightings with Western BFIs within

tual breach by Islamic BFIs when managing Profit Sharing Investment Ac-

the External Level IFRB framework.

counts (PSIAs)); (ii) mark-up risk (risk of losses from fixed mark-up rate
used in contract (e.g., Murabaha) owing to a temporal change in relevant

The RBRs would operate at both an Islamic finance product level and

benchmark rate); and (iii) displaced commercial risk (risk of losses which

an overall institutional level. This would allow Islamic BFIs to implement

Islamic BFIs must absorb to ensure PSIAs are paid rate of return equivalent

appropriate responses, if necessary, to lower RBRs for either a specific

to competitive rate of return) [Toumi and Viviani (2010)].

Islamic product range or overall. Naturally, the higher the number of participating Islamic BFIs, the more accurate the RBRs (in terms of measuring

Consequently, within the current context it is proposed that establishing

relative risk) would be. However, the most benefits would be achieved via

an IFRB is the first and necessary step towards empirically proving this

the implementation of the final External Level IFRB framework. This is be-

claim of Islamic superiority right. Otherwise Islamic BFIs might claim their

cause this framework envisages the use of at least six Islamic BFIs and six

Shariah products to be less risky than Western counterparts owing to

Western BFIs, with ideally double that number, i.e., a syndicate of at least

asset-based financing methods, but in reality may fall short of the mark in

24 BFIs. Western BFIs could be comprised of either those Western BFIs

terms of the overall risk profile offered. For instance, Muslim consumers

with Islamic windows or established stand alone BFIs. The optimal operat-

may purchase a Shariah compliant product which ostensibly shuns riba

ing framework for both levels would be the implementation of public risk

and is asset-based, but which in reality may be far riskier and more prone

tables for a much larger number of Islamic and Western BFIs, as this would

to gharar and maisir than comparative Western financial products. Can

increase the transparency and relative accuracy of risk rankings. The use

it really be said in such instance that this Islamic product complies with

of the ISI as an intermediary between all the participating BFIs would fa-

the spirit of Shariah? The point being made is that Islamic risk bench-

cilitate the acquisition of data needed to determine the RBRs, but without

marking would afford Islamic BFIs an opportunity to realistically verify

compromising the confidentiality and integrity of the information disclosed

prior theoretical claims about Islamic finance products. Indeed, in light

by the participating BFIs. Thus confidential and proprietary information ob-

of Shariah principles, such claims ought to be justified by Islamic BFIs

tained from the BFIs would be protected and not disclosed to any other

if at all possible. Consequently, by providing public and transparent risk

participating BFIs, but the ISI would be authorized to monitor and verify

benchmarkings, Islamic BFIs would take one step closer to objective jus-

the data obtained from the BFIs. In principle the External Level framework

tification of Islamic finance products. This is because if Islamic RBRs are

would offer three significant advantages, namely it would: (i) contribute to

demonstrated to actually be lower in practice than comparative Western

enhancing the legitimacy of Islamic finance; (ii) enhance risk management

RBRs, this would significantly enhance the legitimacy of Islamic finance

systems and practices in Islamic BFIs going forwards; and (iii) afford a new

to Muslim and other consumers. Moreover, even if they are not (per-

opportunity for offering new Shariah compliant Islamic finance products.

haps because of economies of scale or lack of established experience),

The advantages offered by the External Level framework will therefore be

it would demonstrate that Islamic BFIs can still work towards reducing

touched upon, followed by a discussion of an operational road map for

risks for Islamic finance consumers. The underlying Shariah ethos here

implementing the IFRB frameworks in theory.

would be public disclosure and transparency which would also enhance


the legitimacy of Islamic finance. Indeed it can be argued that Islamic

28

finance should not be driven by ruthless pricing or competitive strategies,

The Capco Institute Journal of Financial Transformation


Enhancing Islamic Finance through Risk Benchmarking

82

Self assessments

software (44% did not) [Deloitte (2011)]. In addition, many of those used

6 6 6

70

12 6 12

67

16

11 6

17

18

established risk measures such as VaR (Figure 8), scenario analysis and
industry benchmarks (Figure 9) [Deloitte (2011)]. In light of these findings

Scenario analysts

it is submitted that Islamic risk benchmarking would significantly contrib-

Stress-test
65

Key risk indicators

53

24

ute to the more cohesive development of Islamic risk management best

11 12

practices. It is likely to be the case that Islamic BFIs would strive to lower

Industry benchmark/loss experience


50

19

13

19

RBRs to compete with Western BFIs and in doing so, would continue to

Economic metrics
44

Probabilistic analysis

13

38

Third party assessments


12

Failure mode and effects analysis


0

19

20

19

13
6

13

40

30

improve operational, market, and credit risk identification, mitigation, and


management frameworks. This would likely lead to increased use of risk

37

management software and quantitative risk analysis methods by Islamic

55

60

80

100

BFIs, and possibly the development of more standardized risk measures

120

and modeling techniques. The IFRB could therefore indirectly serve as a

Currently in use

catalyst for Islamic BFIs to update and enhance their overall financial risk

Plan to use
Plan to incorporate in next 12 months

management systems, and the collaborative nature of the benchmarking

No plans to incorporate

system would help Islamic BFIs in mutually enhancing such systems.


Consequently, it is highly likely that these types of developments would

Figure 8 Kinds of risk assessment methods and methodologies used for


risk analysis (%) [Deloitte (2011), 17]

Value at risk (VAR)

42%
37%

Cash flow at risk

also contribute to enhancing the legitimacy of Islamic finance products


in the long run.

Potential new Shariah compliant Islamic finance products


Under Shariah law, riba literally means excess, and in practice refers
to any fixed or pre-determined rates or charges made for the lending of

32%

NPW/ IRR
21%

Economic value added (EVA)


RAROC

11%

Other

11%

money or tied to the maturity or underlying principal sum. For Al-Qaradawi (2001, 54), riba refers to every loan given with pre-conditioned
benefit. The Shariah prohibition therefore expressly condemns the
guaranteed nature of the charge which uses money to make money re-

0%

5% 10% 15% 20% 25% 30% 35% 40% 45%

gardless of the investment performance, or risk undertaken. In addition,


Tabari (2010) believes that the riba prohibition requires that any return
should be accompanied by undertaking a level of risk and liability, and

Figure 9 Kinds of risk measures used (%) [Deloitte (2011), 18]

that the gharar prohibition forbids contracting under conditions of excessive uncertainty and unacceptable levels of risk. Therefore, Western BFI
bank accounts which pay a pre-determined rate of interest on principal

but instead by holistic, equitable and fair risk-sharing initiatives, and the

regardless of risk fall foul of the riba prohibition. However, what about a

IFRB framework would certainly be a step in the right direction towards

Shariah investment of principal with a rate of return linked to the Islamic

achieving this for Islamic BFIs.

BFI risk benchmark? This is not a pre-determined, pre-conditioned, or


guaranteed rate, and what is more, the investment is tied to the underly-

Enhancing risk management in Islamic BFIs

ing benchmark risk of the Islamic BFI, which has historical risk rates and

In a survey conducted on 40 Islamic BFIs (Bahrain, Lebanon, Saudi Ara-

would therefore not comprise excessive uncertainty or levels of risk (i.e.,

bia, Qatar, UAE), 61% of industry leaders believed risk management re-

envisaged low historic volatility rates). In essence, this would therefore

quired new regulatory measures to ensure compliance and best practice,

seek to emulate Western interest bearing accounts but to all extents and

and 66% believed the Islamic finance industry to be under-regulated

purposes structured within the confines of Shariah.

[Deloitte (2010)]. What is more, 50% of those surveyed did not have a
risk management system to address Islamic financial product require-

An alternative potentially Shariah compliant product would be Shariah

ments, and 63.4% agreed that Islamic BFIs were lagging behind in terms

investment certificates tied to an investment in a Shariah investment fund

of implementing risk management systems [Deloitte (2010)]. In a later

(structured under a Mudaraba contract), which linked the rate of return

study involving 20 Islamic BFIs across nine countries (with combined as-

to the IFRB. Alternatively, in recent times the Islamic Waad meaning

sets of more than US$50 billion) 67% of those surveyed used quantita-

unilateral promise, pledge, or firm intention, has been used to structure

tive risk analysis methods (33% did not) and 56% had risk management

a number of innovative developments in Islamic finance derivatives such

29

as forex options which have been used solely for hedging (and not spec-

Stage 1

ulative) purposes. The Waad structure has also been used to arrange

Stage 1 would consist of Islamic and Western BFIs agreeing the blueprint

Islamic finance total return swaps which under conventional Western

for an actual IFRB operating framework. This blueprint would be imple-

structures allow parties to swap payments based on a fixed reference

mented using the 18 financial benchmarking principles contained within

or interest rate or variable, for payments based on returns based on an

the IOSCO Principles, as this would ensure consistency and robustness

underlying asset or index (e.g., listed shares or stock market indexes).

in designing and implementing the IFRB framework across all participat-

Consequently Waad total return swaps might in principle be used to

ing BFIs. In relation to the benchmark design, the IOSCO Principles pro-

gain exposure to Islamic finance products that are not actually owned,

vide at a general level that this should take into account generic design

since the variable leg of the total return swap could be based on the

factors that are intended to result in a reliable representation of the eco-

RBR of a particular Islamic finance product [Uberoi et al. (2009); Attalah

nomic realities of the Interest the Benchmarks seeks to measure and to

and Ghoul (2011)]. Also, Shariah futures and options might in the future

eliminate factors that might result in a distortion of the price, rate, index,

potentially be developed by ensuring that they were structured for hedg-

or value of the benchmark [IOSCO (2013), 8]. At a more specific level,

ing purposes only, based on halal underlying assets, disavowed mutual

they state that relevant factors may include: (i) adequacy of Interest rep-

deferment without actual asset transfer, and used the Islamic product

resentative sample; (ii) size and liquidity of relevant market; (iii) underlying

RBR to eschew avertable uncertainty (gharar) (e.g., by linking the exer-

market size relative to trading volume of reference market; (iv) distribution

cise of an option to only within an accepted specific risk range so that the

of trading among market participants; and (v) market dynamics [IOSCO

contract is not one of pure uncertainty) [Jobst (2007)].

(2013), 17]. These factors would therefore be applied in developing the


IFRB design wherever relevant. The ISI would be responsible for objec-

The IFRB in practice

tively testing the overall design as well as overall benchmark compliance

Firstly, as noted by Allen et al. (2003), in reality all sources of risk (and

of individual participating BFIs.

volatility) are integrated (e.g., credit risk may drive market fluctuations),
and thus models such as VaR should ideally aim to include known cor-

Stage 2

relations among market, credit, and operational risk, wherever possible.

Stage 2 would consist of the participating BFIs reaching a consensus on

There are certainly a number of operational models which have been

the actual risk weightings to be applied across risk typologies, as well as

developed in practice which aim to integrate different risk types (e.g.,

the accepted types of risk modeling techniques. Thus all external and in-

copula, VcV matrix approach, and Bayesian Markov Chain Monte Carlo

ternal risk weightings would have to be mutually agreed in advance. For

simulation) [Jarrow and Turnball (2000), Li et al. (2012)]. However, for

example, an external overall risk weighting for all BFIs of 40:40:20

Islamic finance risk benchmarking it is submitted that while the simple

might be 40% (credit risk), 40% (market risk), and 20% (operational risk).

summation approach (i.e., addition of individual VaR risk measures) to

Alternatively an internal market risk risk weighting of 40:40:20 for an

risk components may ignore diversification benefits, in practice it is to

Islamic BFI might comprise 40% (commodity risk), 40% (equity risk), and

be initially preferred. This is because of the nascent characteristics of Is-

20% (liquidity risk), of a 40% overall market risk share. It is submitted that

lamic finance markets, the complexity of risk identification within Islamic


finance products, together with the conservative nature of this approach
which fittingly reflects the ethos of Shariah. It is further envisaged that as
Islamic and Western BFIs became better acquainted and knowledgeable
about the risk benchmarking framework in practice, operational models
could be developed that more accurately reflected the integration of different types of risk. This would therefore increase the accuracy of the
Islamic risk ratings in the long term. It should also be noted that it is not
intended here to provide a high technical level blueprint comprehensively

Distinct features of Islamic banks compared to conventional


banks
Conventional banks

1. Shariah compliance

Non-existent

1. Shariah risk

2. Prohibition of riba

Based on interest rates

2. Rate-of-return risk
3. Mark-up benchmark risk

3. Lending facilities must be


backed by a physical asset
(PLS or sale-based)

Facilities are only based on


lending money based on
interest rates

4. Commodity price risk


5. Increase operational risks for
delivering/holding asset

4. Having PLS contracts

Non-existent

6. Equity investment risk


7. Increase operational risk (and
asymmetric information)

5. Restrictions in requesting
collaterals and penalties

No restrictions imposed

8. Increase credit risk

6. Investment accounts
(deposits) are based on a
mudaraba contract

All deposits are determined by


interest rates

9. Displaced commercial risk


(DCR)

7. Restrictions on secondary
markets and interbank
activities

Secondary markets witness


continuous innovations

10. Increase liquidity risk

focusing on methodological issues. This is something that could be developed in time using internal Islamic and Western BFI knowledge of relevant market, credit, and operational risks, and collaboration among IFRB
participants. Instead, the theoretical possibility of developing a comprehensive IFRB framework will be addressed, including the identification of
risks specific to different types of Islamic finance contracts, together with
other relevant operational issues. It is intended that the IFRB framework
30

be implemented via a three-stage process.

Changes to Islamic banks risk


profile

Islamic banks

Figure 10: Changes to the risk profile caused by the distinct features of
Islamic banks [Salem (2013), 45]

The Capco Institute Journal of Financial Transformation


Enhancing Islamic Finance through Risk Benchmarking

Payment instalments
=

Credit
liquidity risk

Maturity date

Cycle angle payment

Credit risks

Operational
risk

Price + profit

Payment inability

Balance sheet
P and L
fluctuations

Market price

Profit {

Ownership and promise

Operational risk

Damage of the asset

Credit risks

Maturity
date

Delayed
payments

Default of payment

Major catastrophe

Earlier leave

Payment rent inability


1

PF

Receiving commodity

Variable payment
instalments

Liquidity risk

Order for purchasing

Market risk

Received back

4
Loss

Contract deal date

Buy in advance and lease

Fixed payment

Sell the asset

Market risk

Commodity price fluctuations

Market price
fluctuation

1
Commodity price risk

Market risk

Re-sell the commodity

Market risk

Commodity price risk


Balance sheet
fluctuations
P and L
fluctuations

In case of default collateral and


guarantees maybe applied
Future cash flows on delayed
payments

Figure 11: Credit, operational, and market risks during the lifetime of
Ijarah contracts [Akkizidis and Khandelwal (2008), 71]

Figure 12: Credit, operational, and market risks during the lifetime of a
Murabaha contract [Akkizidis and Khandelwal (2008), 56]

this stage would likely be the most complex as well as controversial stage

Islamic bank risks should be identified on two levels?

of the process. This is because some may argue that Islamic banking risks

It is proposed that in theory Islamic bank risks should be identified on two

are unique and cannot be compared or modeled to comparative Western

levels, the first being the overall bank level where the sources of risk are

banking risks. However, it is opined here that this line of argument is weak,

analyzed (i.e., market, credit, operational risks, and weightings), and the

because although risk profiles may differ, overall risk exposure can be cal-

second level being specific risks arising in each of the main Islamic finance

culated and compared, and this is what matters. The existing financial risk

contracts [Salem (2013)]. For instance, Salem (2013) remarks that market

is very real, and therefore what matters is how precisely such risk can be

risk in Islamic BFIs comprises mark-up, commodity price, forex, and equity

identified, quantified, and modeled in practice. For instance, two cars on

risk, where the first two are specific to Islamic finance contracts, but the

a motorway may have overall risk exposures that differ according to the

last two are identical for conventional banks. It is further noted that opera-

different inherent risk profiles (e.g., age and experience of driver; drivers

tional differences can be identified by examining the particular operating

appetite for speed; car build and safety level; age of car), but they can

models and the mapping of different risks, resulting in changes to the risk

still be compared fairly accurately according to different risk factors and

profile of Islamic BFIs (Figure 10). Consequently in practice this mapping

weightings in order to highlight relative collision risk.

could therefore be used by Western and Islamic BFIs to mutually agree


comparative risk weightings for external overall and internal specific risk

Stage 3

weightings. For example, it might be agreed that all IFRB BFIs might have

Stage 3 would consist of actual implementation of the IFRB framework

a 20% operational risk weighting, but internal weightings within this 20%

and subsequent RBR monitoring by the ISI. It would include scheduled

operational risk might differ. This is because Islamic BFI internal risk weight-

reviews of the operating framework with a view to increasing the reliability

ings might instead include Shariah risk, displaced commercial risk, and

and accuracy of the RBRs. In order to maximize the accuracy of the IFRB

mark-up risk. This would require an ex ante analysis of risk typologies for

it is envisaged that there would be a six month test phase during which

each Islamic finance contract offered by relevant Islamic BFIs. For example,

the accuracy of the IFRB could be tested prior to the final official rolling

Ijara contracts involve leasing an asset (commodity price risk) together with

out phase. Consequently, it is suggested that during the first six month

fixed rentals (mark-up risk) (Figure 11), whereas in Murabaha contracts no

test phase of operations, weekly scheduled reviews of the operating

commodity price risk arises (asset acts as collateral) but market risk arises

framework could be undertaken, changing to monthly scheduled reviews

in the shape of price risk (i.e., mark-up risk) (Figure 12) [Salem (2013)]. The

thereafter. In practice it is envisaged that the ISI would function in a simi-

overall idea would therefore be to map the particular complexities in Islamic

lar fashion to the way credit rating agencies such as Fitch, Moodys,

finance products, and for all Islamic and Western BFIs to agree an equiva-

and Standard & Poors work. In theory the historical data obtained by

lent risk measurement for comparative products and sectors.

the ISI could also help to improve the accuracy of financial risk modeling
within Islamic BFIs, although disclosure of such information would ne-

It is acknowledged that there would be many complications arising from

cessitate prior absolute consent and agreement by all participating BFIs.

the rolling out of the risk benchmarking process, but at the same time it is

31

2
P and L
fluctuations

Market risk

Market risk

Level of investment
Share price fluctuations

Market price
Last equity
payment and/or
sum of payments

Payment investment and


covering losses

Initial investment + profit

Loss

Payment instalments

Profit

Cycle angle selling


shares

Cycle angle income

Operational risk
Credit risk
Liquidity risk

Credit risk
Liquidity risk

Payment inability

Major loss

Inability to carry on
business

Operational risk
Business risk

Balance sheet
fluctuations

End of partnership

Loss

Loss

Investment principle

Profit

1
2

Figure 13 Credit, operational, market, and liquidity risks during the


lifetime of Permanent Musharakah contracts [Akkizidis and Khandelwal
(2008), 45]

Figure 14 Credit, operational, market, and liquidity risks during the


lifetime of Diminishing Musharakah contracts [Akkizidis and Khandelwal
(2008), 46]

opined that the process is realistic and can be achieved. In fact there are

finance contracts can be mapped out in terms of relevant credit, opera-

two main difficulties that may arise in this type of risk modeling. The first

tional, market, and liquidity risks, in practice translates to mean appropri-

is that this type of Islamic contract risk measurement would be required

ate Islamic finance risk models could be developed and applied. Moreover,

to be dynamic in nature, since the Islamic finance contract structure may

it also means that relative risk weightings and a risk hierarchy could also

change over time, and static risk measurements would fail to capture

be deduced. For example, Salem (2013) concludes that profit-loss-shar-

the verity of the underlying transaction. For instance, for a Musharakah

ing instruments such as the Mudaraba pose the highest risk, followed by

contract, risk weightings for credit, operational, market, and liquidity risks

product-deferred instruments such as the istisnaa (custom manufactured

may change over time depending on whether it is permanent (Figure 13)

goods deferred delivery contract), and then lowest risk Ijara and Mura-

or diminishing (Figure 14) in nature. The second difficulty is that the role

baha (Figure 15 and Figure 16) (a more fixed and predictable margin). Con-

of information management for Islamic finance BFIs is more critical as

sequently, all that is then needed is the development and application of

compared to Western BFIs, as Islamic contracts typically require more in-

relevant risk modeling methodologies, which will in turn depend on the

tegration with the activities of the entrepreneur, meaning there is a heavy

availability of relevant information. Modern Western and Islamic BFIs are

bias towards information availability (e.g., in Musharakah and Mudarabah

more than capable of developing and applying these types of risk model-

contracts) [Akkizidis and Khandelwal (2008)]. This is one reason in favor

ing methodologies. For instance, BFIs full historical loss data for specific

of the prior implementation of the interim Internal IFRB framework, as it

partnership agreements such as Musharaka and Mudaraba would allow

would afford Islamic BFIs an opportunity to enhance their information

historical simulation or VcV approaches for calculating credit VaR, where-

management capabilities, and to fine tune their risk management operat-

as partial or complete lack of historical loss data would instead necessi-

ing frameworks. Indeed it can be argued that transparency of information

tate Monte Carlo simulation estimates [Akkizidis and Khandelwal (2008)].

is crucial for designing an effective IFRB framework, and is essential for

Therefore much would depend on the existing information available to the

the accurate measurement and modeling of risks (e.g., information used

BFI, as well as the strength and robustness of the BFIs existing risk man-

for defining stop loss sale value, or in static and dynamic analysis for esti-

agement framework. It is therefore envisaged that the IFRB benchmarking

mating current and future VaR for market risk in Musharakah, Mudarabah,

framework would have mutually accepted market, credit, operational, and

and Ijara contracts) [Akkizidis and Khandelwal (2008)].

liquidity risk measurements systems in place for all participating BFIs. The
importance of such systems would lie, not in disclosing the proprietary

32

Nevertheless, the fact that the functionality of the different types of Islamic

nature of the systems, but in ensuring that their qualitative or quantitative

The Capco Institute Journal of Financial Transformation


Enhancing Islamic Finance through Risk Benchmarking

Payment investment and


covering losses

P and L
fluctuations

Market risk
Level of investment

Last equity
payment and/or
sum of payments

Payment investment and


covering losses

Cycle angle income

Inability to carry on
business investment

Investment principle

Liquidity risk

Credit risk
Liquidity risk

2
1

Figure 15 Risks in Mudarabah contract agreements during the


investment period [Akkizidis and Khandelwal (2008), 51]

Balance sheet
fluctuations

Inability to carry on
business

Operational risk
Business risk

Major loss

Operational risk

Loss

Cycle of investments

Major loss

Loss
Business risk

Figure 16 Risks in Mudarabah contract agreements during the business


profitability and loss period [Akkizidis and Khandelwal (2008), 52]

validation, functionality, back-testing, and stress-testing was undertaken in

Other arguments that have arisen in relation to Islamic finance include

a manner supervised and approved by the ISI in accordance with the es-

that the Islamic finance industry is under-regulated, it lacks global cohe-

tablished IOSCO benchmarking standards. In this way, the External Level

sion and uniformity, and owing to its somewhat restrictive nature it has

IFRB framework and RBR monitoring by the ISI could be developed in a

difficulties in developing new Shariah compliant products and services.

cohesive and realistic manner over time.

At the same time, arguments have been made that owing to the ethical
and equitable framework of Islamic finance, Islamic finance contracts are

Conclusion

safer than Western counterparts (in terms of financial risk), since on the

It has been seen that Islamic finance is a globally established method

whole they are finance contracts that are based on identifiable underly-

of financing that conforms to the traditions and requirements of the Is-

ing assets. The limited financial losses incurred by Islamic BFIs during

lamic faith. In practice this means that Islamic BFIs must provide finan-

the recent financial crisis would seemingly testify to such a notion. Nev-

cial services, transactions, or contracts that conform with Shariah, that

ertheless, investor losses experienced through the default of the UAEs

is, the rules making up Islamic commercial jurisprudence. These include

Nakheel sukuk in 2008-2009 owing to the Dubai property crash, also bear

adhering to the prohibitions against riba (interest), gharar (uncertainty),

testament to the fact that Islamic finance is not immune to large financial

and maisir or qimr (gambling and speculation), as well as ensuring that

losses. In light of these propositions it has been argued here that the de-

any products or services provided or involved in Islamic finance are not

velopment of a new collaborative risk benchmarking framework for pub-

haram (i.e., forbidden). In addition, all aspects of financing undertaken by

lic and private Islamic and Western BFIs would contribute to significantly

Islamic BFIs are monitored by relevant SSBs in order to ensure confor-

ensuring Islamic finance going forwards. Most notably it could contribute

mity with Shariah principles and practices. In light of these requirements

to enhancing the legitimacy of Islamic finance products, it could enhance

Islamic finance has developed a large number of financial contracts that

risk management systems and practices in Islamic BFIs, and it could pro-

have been established as conforming to the requirements of Shariah, in-

vide new opportunities for the development of new Shariah compliant

cluding, but not limited to, the Murabaha, the Mudaraba, the Musharaka,

Islamic finance products. It has been seen that the External Level IFRB

and the ijara contracts, as well as Islamic sukuk. Nevertheless, while Is-

framework would map out all the most relevant market, credit, opera-

lamic finance has become a globally established stream of finance, it is

tional, and compliance or Shariah risks. It would also utilize the IOSCO

still undergoing further development. For example, it has been seen that

Principles to provide a robust overarching framework for benchmarking

many Islamic BFIs may lack comprehensive risk management systems.

operational RBRs for all participating BFIs. From one perspective it could

33

be argued that this type of collaborative benchmarking would drive value,

enable continuous risk management improvement, and would lend transparency to Islamic finance operations. It might also be argued that the
External Level IFRB framework could galvanize participating Islamic BFIs

into action vis--vis risk management capabilities, and would also significantly contribute to overcoming the paradigm blindness regarding the
need for competitive advantage at all costs. Consequently in light of the
underlying ethos and principles of Shariah, it is submitted that the collaborative benchmarking of risk of Islamic and Western BFIs could lead
to significant new benefits for all participating BFIs across-the-board.

References

Abdulkader, T., S. Cox and B. Kraty, 2005, Structuring Islamic Finance Transactions, London
Euromoney Institutional Investor PLC

Ahmad, A., 1993, Contemporary Practices of Islamic Financing Techniques, Jeddah: Islamic
Research and Training Institute, Islamic Development Bank
Akkizidis, I. and S. K. Khandelwal, 2008, Financial Risk Management for Islamic Banking and
Finance, Hampshire: Palgrave Macmillan
Allen, L., J. Boudoukh, and A. Saunders, 2003, Understanding Market, Credit and Operational
Risk: The Value at Risk Approach, Oxford: Blackwell Publishing
Allen, S. L., 2013, Financial Risk Management: A Practitioners Guide to Managing Market and
Credit Risk, Chichester: John Wiley & Sons
Al-Qaradawi, 2001, Banks Interest is the Prohibited Riba, Cairo: Wahba Publishers
Anand, G. and R. Kodali, 2008, Benchmarking the benchmarking models, Benchmarking:
An International Journal, 15(3): 257-291
Arora, N., J. R. Bohn, and F. Zhu, 2005, Reduced Form vs. Structural Models of Credit Risk:
A Case Study of Three Models, Moodys KMV Company: 1-39
Attalah, C. C. and W. A. Ghoul, 2011, The Wad-Based Total Return Swap: Sharia Compliant or
Not?, The Journal of Derivatives, 19(2): 71-89
Ayub, M., 2008, Understanding Islamic Finance, Chichester: John Wiley & Sons
Bank for International Settlements (BIS), 2006, Basel II: International Convergence of Capital
Measurement and Capital Standards: A Revised Framework - Comprehensive Version, Bank
for International Settlements (June)
Black, F. and M. Scholes, 1973, The Pricing of Options and Corporate Liabilities, Journal of
Political Economy, 81: 673-654
Buehler, K., A. Freeman and R. Hulme, 2008, Owning the Right Risks, Harvard Business
Review (September) 1-9
Camp, R. C., 1989, Benchmarking: The Search for Industry Best Practices that Lead to
Superior Performance, Milwaukee: ASQC Quality Press
Chinoy, S, 1995, Interest-free banking: the legal aspects of Islamic financial institutions,
Journal of International Banking Law, 10(12): 517-524
Deloitte, 2010, The Deloitte Islamic Finance leaders survey in the Middle East Benchmarking
practices, Biannual Survey (Issue 1)
Deloitte, 2011, Empowering Risk Intelligence in Islamic Finance, Managing risk in uncertain
times, Deloitte in the Middle East
DeLorenzo, S.Y.T., 2007, Symposium: Islamic Business and Commercial Law: Shariah
Compliance Risk, Chicago Journal of International Law, 7(2): 397-408
Delpachitra, S. and D. Beal, 2002, Process benchmarking: an application to lending products,
Benchmarking: An International Journal, 9(4): 409-420
Ernst & Young, 2012, Global Islamic assets are expected to reach $1.8 trillion by 2013 (10
December, Dubai), available at: http://www.ey.com/EM/en/Newsroom/News-releases/10-Dec2012Global-Islamic-assets-are-expected-to-reach---1-8-trillion-by-2013
Fadeel, A., 2002, Legal aspects of Islamic finance, in Karim, R. A. A. and S. Archer (eds), Islamic
Finance: Innovation and Growth, London: Euromoney Books
Fong, S. W., Cheng, E. W. L., and Ho, D. C. K., 1998, Benchmarking: a general reading for
management practitioners, Management Decision, 36(6): 407-418
Gerken, A., N. Hoffmann, A. Kremer, U. Stegemann, and G. Vigo, 2010, Getting risk ownership
right McKinsey Working Papers on Risk, 23: 1-13
Hanif, A., 2008, Islamic finance an overview, International Energy Law Review, 1: 9-15
Hull, J. C. and A. White, 2000, Valuing Credit Default Swaps I: No Counterparty Default Risk,
(April) NYU Working Paper No. FIN-00-021
Ioannis, A., K. Lampros, Z. Ioanna and B. Vivianne, 2006, Integrating Market, Credit and
Operational Risk: A Complete Guide for Bankers and Risk Professionals, London: Risk Books

34

International Organization of Securities Commissions (IOSCO), 2013, Principles for Financial


Benchmarks: Consultation Report, The Board of the International Organization of Securities
Commissions, i-v: 1-37
Jabbar, H. S. F. A., 2009, Islamic finance: fundamental principles and key financial institutions,
Company Lawyer, 30(1): 23-32
Jarrow, R. A., and S. M. Turnball, 1995, Pricing Derivatives on Financial Securities Subject to
Default Risk, Journal of Finance, 50: 53-86
Jarrow, R. A. and S. M. Turnball, 2000, The intersection of market and credit risk, Journal of
Banking & Finance, 24: 271-299
Jobst, A. A., 2007, Derivatives in Islamic finance, Islamic Economic Studies, 15(1): 1-32
Khnifer, M., 2010a, When Sukuk Default Asset Priority of Certificate-Holders Vis a Vis
Creditors, Opalesque Islamic Finance Intelligence, 11(9)
Khnifer, M., 2010b, When Sukuk Default Certificate Holders Can Take Default Worries Off
Their Shoulders with Proper Asset-Backed Sukuk, Business Islamica Magazine: 20-25
Li, J., J. Feng, D. Wu, C-F. Lee, 2012, Integration of Credit, Market and Operational Risk:
A Comparative Analysis of Copula and Variance/Covariance Approach, Department of Finance
and Economics, Rutgers Business School, Rutgers University
Lipman, H., 2013, Regulating Financial Benchmarks: Creating the Right Balance, Bloomberg
Blog available at: http://blog.bloomberg.com/2013-03-04/regulating-financial-benchmarkscreating-the-right-balance/
Mainelli, M., 1999, Taking the Measure of Risk: Benchmarking Risk Management, Handbook
of Risk Management, 35: 5-8
McNair, C. J. and K.J.H. Leibfried, 1992, Benchmarking, New York: Harper Collins
Medova, E. A. and M.N. Kyriacou, 2001, Integrating Market, Credit and Operational Risk:
A Theoretical Framework for Valuation Integrity, Presentation for Risk 2001 Europe (10-11
April), Centre for Financial Research, Judge Institute of Management Studies, University of
Cambridge
Merton, R. C., 1974, On the pricing of corporate debt: The risk structure of interest rates,
Journal of Finance, 29: 449-470
Merton, R. C., 1977, On the pricing of contingent claims and the Modigliani-Miller theorem,
Journal of Financial Economics, 5: 241-249
Mohammed, A. and S. Kianfar, 2007, School of Sharia, European Lawyer, 66: 32-34
Moriarty, J. P., 2011, A theory of benchmarking, Benchmarking: An International Journal,
18(4): 588-612
Mukherjee, A., P. Nath and M.N. Pal, 2002, Performance benchmarking and strategic
homogeneity of Indian banks, International Journal of Marketing, 20(3): 122-139
PricewaterhouseCoopers, 2011, Raising the bar for risk management, Benchmarking risk
management practices in the banking industry in EMEAI, December: 1-8
Rethel, L, 2011, Whose legitimacy? Islamic finance and the global financial order, Review of
International Political Economy, 18(1): 75-98
Salem, R.A., 2013, Risk Management for Islamic Banks, Edinburgh: Edinburgh University Press
Schoon, N., 2008, Islamic finance an overview, European Business Organization Law
Review, 9(4): 621-635
Smithson, C. and P. Song, 2004, Quantifying operational risk, Risk (July): 57-59
Sol, J., 2007, Introducing Islamic Banks into Conventional Banking Systems, IMF Working
Paper WP/07/175: 1-26
Standard and Poors, 2013, Risk-to-Price: The New Risk Benchmark, available at: http://
www.standardandpoors.com/products-services/risk-to-price/en/us
The Banker, 2008, Top 500 Islamic financial institutions, November
Toumi, K. and J. L. Viviani, 2010, Islamic banks exposure to displaced commercial risk:
Identification and measurement, CR2M, Universit Montpellier 1, 1-23
Tucker, F. G., S.M. Zivan, and R.C. Camp, 1987, How to measure yourself against the best,
Harvard Business Review, 87(1): 2-4
Uberoi, P., R. Chatterji and D. Bidar, 2009, The waad on the Street, Risk Magazine, 1 August
van Greuning, H. and Z. Iqbal, 2008, Risk Analysis for Islamic Banks, Washington: The World
Bank
Wendel, C. B., 1993, 4 sharp tools for learning how your bank measures up, American
Banker, 27 September: 18-19

High-level Debate

Credit Valuation
Adjustment: The
Devil is in the Detail
Sylvain Prado Consultant, Capco
Bhavdeep Virdee Consultant, Capco

Abstract
On September 15, 2008 Lehman Brothers unexpectedly

risk measurement is Credit Valuation Adjustment (CVA).

went bankrupt. It was the fourth largest investment bank in

There is an abundance of papers on CVA. Unfortunately they

the USA at that time. Lehman could no longer honor its debt

are often vague, confusing and over technical. With this pa-

and swap obligations. Financial institutions in transactions

per we aim to demystify CVA and provide a straightforward

with the bank were facing huge counterparty risk with the

perspective on computation of CVA. In Sections 1 and 2, we

bank defaulting on its obligations.

define Credit Valuation Adjustment and explain why it mat-

Following that month we have seen the biggest financial cri-

ters. Section 3 decomposes its calculation.

sis since the 1920s. There have been significant changes to

We also aim to contribute to the literature on CVA by bring-

the regulatory landscape across the globe since then. Regu-

ing a new question on the table: Beyond the usual theoreti-

lators including the Federal Reserve, FSA, FINMA and others

cal discussions, what are the real consequences for a bank

made amendments to the way counterparty risk should be

when implementing CVA? Section 4 provides answers and

measured and capitalized. The main element of counterparty

Section 5 concludes.

35

What is credit value adjustment (CVA)?

Basel III requires banks to capitalize counterparty risk.

CVA reflects how much the default-free price of a derivative should be

As a consequence of the 2008 crisis, CVA now constitutes a large part

adjusted to account for the possibility of a counterparty default. In terms

of the regulatory capital framework for the trading book. Banks need a

of capital, it translates to how much one must hold in reserves to cover

CVA risk capital charge to comply with the latest Basel III requirements

the risk of changes in the mark to market (MtM) values of over the counter

[BIS (2011)]: In addition to the default risk capital requirements for coun-

(OTC) derivatives.1 It is worth noting that almost two-thirds of losses that

terparty credit risk determined based on the standardized or internal

occurred during the financial crisis arose from MtM movements rather

ratings-based (IRB) approaches for credit risk, a bank must add a capital

than outright defaults. Financial institutions must estimate CVA to price

charge to cover the risk of mark-to-market losses on the expected coun-

counterparty credit risk into their OTC derivative contracts.

terparty risk to OTC derivatives.

CVA should never be considered in isolation. For instance, from a pric-

CVA capital will be integrated with other components of trading book

ing perspective it is a component of the theoretical value of a derivative.

capital. The approvals obtained on exposure modeling, the use of a stan-

This is made up of the default free value, the debit valuation adjustment

dardized or an Internal Model Method (IMM) will impact the approach to

and the funding valuation adjustment. From a capital perspective, CVA is

model CVA regulatory capital.2

linked to other components of trading book capital such as market risk,


liquidity risk, and operational risk.

In parallel the Basel Committee recommends an independent review of


counterparty risk management at regular intervals (not less than one a year).

Why does it matter?

As reported in Table 1 (p. 24), different perspectives are under consideration.

CVA enables active management of counterparty credit risk.


CVA enables desks to quantify counterparty risk through pricing adjust-

What are the building blocks of a CVA calculation?

ments and understand how counterparties contribute to the overall risk

Calculating CVA requires the understanding of the pricing of a deriva-

of a portfolio. As explained below, the risk must be embedded into the

tive product. An example with an interest rate swap describes below the

deal assessment and the pricing process.

components of CVA.

CVA allows pre-trade pricing and product structuring with interac-

CVA is always calculated at the trade level.

tion between different departments in a bank.

First we use market data to calibrate an interest rate model. Then we

The market price of counterparty risk provides a better valuation of OTC.

can apply Monte Carlo simulation to generate various scenarios for the

Traders need to report CVA charges of the deals through an accurate

floating interest rate. The estimation of the Expected Exposure (EE)3, de-

calculation. CVA enables banks to reflect the price of counterparty risk in

fined as the average of the distribution of exposure at any future date,

new trades on a real-time basis. As a result, daily and intra-day calcula-

constitutes a critical step.4 We have to include all the deal economics

tion of CVA has become a norm in many banks.

of the swap to generate exposure profiles. Finally, various counterparty


parameters are combined to calculate the CVA.

As illustrated in Figure 1, different parts of a bank will view the market


risk differently. These views must be reconciled during the development

Different modeling approaches can be applied, depending on the nature

of the CVA framework.

of the portfolio and required level of sophistication [Brigo et al. (2007),


Zhu and Pykhtin (2007)].

CVA enables the bank to decide on the best way to manage exposure and to choose the appropriate hedging strategy.

Figure 2 illustrates the structure of the cash flow for an interest rate swap.

Banks buy credit protection to mitigate counterparty risk. Multiple ques-

Figure 3 illustrates the steps to compute CVA for the interest rate swap.

tions, however, need to be answered: At what price should the protection be bought? How much credit protection should be bought? What is
the right time to re-hedge the underlying counterparty risk? Can we link
credit and market risks and then hedge the combined risk? How does a
trader manage wrong way risk or the correlation between the exposure
and default probability of an obligor?
These questions should be addressed by the bank and take into account
36

the perspective of each department.

1 As a consequence, OTCs only are subject to CVA.


2 See section 3.
3 The recovery rate, the discount factor and the probability of default are fundamental
elements of counterparty credit risk and vast literature already exists with large number of
tools available for practitioners.
4 Expressing CVA as a mathematical equation we have CVA (at time Tn) = (1-Recovery Rate)
* Discount Factor * Expected Exposure * Probability of Default

The Capco Institute Journal of Financial Transformation


Credit Valuation Adjustment: The Devil is in the Detail

Credit Risk Department


Manage and monitor the
risky MtM value

Credit administration
Counterparty credit
exposure and limit
management

Accounting
Accountants need to add
unrealised losses on books

Impact of risky
MtM value on?

Pay fixed @ Z.ZZ%


5-year interest rate
swap with semi-annual
payments

Funding
Traders need to source
additional funds to address
the liquidity deficit

Counterparty A

Counterparty B

Receive floating
@ LIBOR + Y.YY%
with quarterly payments

Liquidity risk
Manage and monitor the
funding gap

Market data
acquisition

Figure 1 Views of mark-to-market risk in different departments of a bank

Figure 2 Structure of the cash flow for an interest rate swap

Calibration to initial
yield curve

Boot strapping

Zero coupon yield curve

Calibrated model
parameter
values are
the inputs

Stochastic model for the


term structure of Interest
rates

Monte Carlo
simulation to
generate spot yields
based on zero
coupon yield

0.06
0.05

Interest rate value

Two methods to perform


Monte Carlo simulations

Model calibration

LIBOR
Market
yieldRate
curve

Parameters
calculated
using
various
methods like
Maximum
Likelihood
Estimation
and Kalman
filtering

0.04
0.03
0.02
0.01

An alternate stochastic model can be used for Monte


Carlo simulations to generate spot yields

0
100
50
0

100

80

60

40

20

Forward curve0

-3

x 10
3.5

Example, Hull-White Method


using Annuity value

Simulation to
generate
exposure profiles

2.5

Expected Exposure

Swap deal economics


(frequency, notional,
duration, swap rate)

Two methodologies to
generate EPE values

Scenario generation

Simulations

1.5

Or benchmark, for example,


Black and Scholes method
using DV01

0.5

0
15

10

100

60

70

80

90

50

30

40

20

10

Periods

Simulations

Note: CVA distribution

Wron
g

way
of sim risk impa
c
ulate
d inte t through
c
rest r
ate to orrelation
PD

CVA distribution

Payer or receiver
perspective

Counterparty information (PD, recovery rate, discount factor)

Figure 3 The steps to compute CVA for the interest rate swap

2.5

Population in Buckets

Pricing and exposure


management

Rating - BBB (S&P)


Recovery rate X.XX%

Estimation of the
CVA and
adjustment of a
new Mark to
Market (MtM)

1.5

0.5

0
2.4

2.6

2.8

3.2

3.4

3.6

3.8

4.2
-4

CVA in % of the
nominal amount

x 10

37

As specified in Basel III, there are three possible combinations in the cal-

Documentation
Adequacy of documentation of counterparty risk management system and process

culation of the Total Counterparty Credit Risk Capital (CCR) charge. And
the formulas developed by regulators are not really intuitive.

People, processes and governance


Organization of collateral management unit
Organization of counterparty risk control unit

For banks with Internal model method (IMM) approval and market-risk

Integration of counterparty risk measures into daily risk management

internal-models approval for the specific interest-rate risk of bonds, the

Approval process for risk pricing models and valuation systems used by front and back-

total capital charge is the sum of the advanced CVA risk capital charge

office personnel
Validation of any significant change in counterparty risk measurement process

and the higher value of either the IMM capital charge based on current

Scope of counterparty credit risks captured by risk measurement model

parameter calibrations for Exposure at Default (EAD) or the IMM capital


charge based on stressed parameters calibrations for EAD.

Data acquisition
Integrity of management information system

Advanced CVA capital calculation

Accuracy and completeness of counterparty risk data


Accurate reflection of legal terms in collateral and netting agreements into exposure
measurements verification of consistency, timeliness and reliability of data sources used
to run internal models, including independence of such data sources

CVA= LGD MKT .

Max 0;exp i=1

Quantitative analysis and statistical modeling


Accuracy and appropriateness of volatility and correlation assumptions
Accuracy of valuation and risk transformation calculations and verification of models

S i-1 .ti-1
S i .ti
-exp LGD MKT
LGD MKT

EE i-1 .D i-1 +EE i .D i


2

Total counterparty credit risk capital aims to capture default, credit


migration and spread risk and is given by:
Total CCR Capital = CCR (default) capital + Advanced CVA risk Capital

accuracy through Frequent back-testing

For banks with IMM approval and without specific-risk VaR approval

Table 1

for bonds, the Total CCR Capital is the sum of the standardized CVA
CVA incorporates portfolio dynamics.

risk capital charge and the higher value between the IMM capital charge

Moving to a portfolio level requires the inclusion of netting and collateral

based on current parameter calibrations for EAD and the IMM capital

adjustments. A netting agreement established contractually, allows the

charge based on stressed parameter calibrations for EAD.

aggregation of transactions between two counterparties. Transactions


with negative value can be used to offset the ones with positive value and

Standardized CVA capital calculation

reduce the total exposure. A margin agreement, also established contractually, requires that a counterparty posts additional collateral when

2
hedge

0.5.wi . Mi .EAD total


-Mi
i

K=2.33. h.

Bi -Wind .Mind .Bind

the exposure exceeds a threshold. When the exposure goes below the

hedge

0.75.wi2 . Mi .EAD total


-Mi
i

Bi

specified threshold then the collateral is returned.

h = 1 (one-year horizon)

Rating

Bi and Bind denote the hedging


positions in single name and indices

AAA

0.7%

AA

0.7%

als granted to the bank by financial authorities. It determines whether a

0.8%

standardized or an advanced CVA risk capital charge (calculated with the

BBB

1.0%

internal models) can be applied.

2.0%

CVA for capital measurement is not the same as CVA for pricing.
The CVA capital charge depends on the capital methodology approv-

B
CCC

Weight wi

3.0%
18.0%

Total CCR Capital = CCR (default) capital + Standardised CVA risk


Capital

Netting

For all other banks, it is the sum of the standardized CVA risk capital
charge and the standardized method based capital charge or the sum
Trade level
CVAs

OTC
Portfolio

Collateral

over all counterparties of the current exposure method.


The formulas above always emphasize the criticality of the exposure pro-

38

Figure 4 Portfolio perspective

file.

The Capco Institute Journal of Financial Transformation


Credit Valuation Adjustment: The Devil is in the Detail

Implementing CVA remains a tough challenge

Banks have to incorporate these improvements in order to manage their

The previous section outlines the steps to calculate CVA from a pric-

modeling of CVA, make best use of capital optimization opportunities

ing and capital perspective. Unfortunately, additional elements must be

and differentiate themselves from competitors. A model inventory is re-

taken into account in order to implement a modelling framework for CVA.

quired to clearly understand the linkages between the various components of CVA.

Data management constitutes a first roadblock.


Data acquisition The ability to bring together trade, market and refer-

Model validation A robust model validation framework mitigates model

ence data across the front office, risk, and finance in a harmonized man-

risk and ensures compliance with the latest regulations. On-going vali-

ner is fundamental. A major requirement must be the effective and timely

dation activities, assessments of the model documentation, appropriate

capture of time series data used to derive the risk metrics.

back-testing, and stress testing of the models must be supervised by risk


methodology teams.

Product taxonomy Payoff and product definitions of the trades need


to be consistent.5 Dodd Frank Act (2010) and the OCC (2011) guidelines

Model development infrastructure The bank must harmonize sce-

on model risk management require financial institutions to develop strong

narios and market risk factors across all products. Systems infrastructure

product taxonomy.

should enable products booked and priced in different systems to utilize


consistent market factors. The appropriate handling of time series calcu-

Statistical modeling raises various concerns.

lations and the use of technology to optimise computation time have to

CVA methodology: key modeling improvements of many industry partici-

be taken into account.

pants are listed in Table 2.


Strategic decisions are impacted by CVA.
The new trading book rules (particularly CVA and systemic risk) might
Credit exposure
modeling

Wrong way risk

The expected positive exposure (EPE) is a main metric.


Many banks are making a large investment in their exposure
modeling capabilities. Depending on the level of sophistication
that a bank wants to achieve, different methodologies can be
applied.
The correlation between exposure and the probability of default
can be incorporated through different techniques (micro level
correlation between stochastic processes, copula models, the
use of expert inputs from traders, signal correlation switches)

Credit spread
calibration for CVA

In the advanced CVA capital calculation, we generally use


the term structure of CDS spreads. As a result dealing with
counterparties without liquid Credit Default Swap (CDS) curves
involves innovations. And banks are using data cubes and
clustering analysis techniques to determine proxies.

Collateral close-out
risk

Some models jointly simulate the derivatives and associated


collateral in order to consider the impact of different close out
periods.

Collateral close-out
contagion

More advanced models consider scenarios where collateral


can effectively dry-up due to several market participants
requiring liquid collateral.

Dual curve (blended)


vs. Overnight
Index Swap (OIS)
discounting

Most banks use LIBOR for discounting uncollateralized


transactions and OIS spreads for collateralized transactions.
A debate remains as to whether one should use a blended rate
or apply classic spreads across the board.

Correlation risk

It means assessing the correlation between risk factors, the


credit migration and the default of counterparties. From a
pricing perspective, banks need to consider the impact of
cross-gamma (sensitivity) between market factors. From a
capital perspective, one can employ ratings based models,
structural models or hazard rate models.

CVA hedging

CVA can be hedged through the use of single name CDS,


index CDS, sovereign CDS and proxy hedges.

Table 2 CVA methodology: key modeling improvements of many


industry participants

reveal that a previously profitable deal may not be so under the new regime. Banks will learn how new levers can be pulled to influence pricing
and capital consumption, thereby achieving the desired balance between
risk and reward.

Conclusion
This article gives the reader an opportunity to clarify, beyond usual theoretical literature, the key considerations for a bank when developing a
CVA framework.
CVA is used for the pricing adjustment of OTC derivatives and as an additional component of reserves calculations. As a consequence, the deal
assessments, the hedging techniques and other strategic decisions of a
bank are strongly impacted.
CVA contains, by definition, a complex blend of credit and market risk
parameters. Given the rapid changes in the regulatory landscape and the
need to be competitive, it is crucial for an organization to create a flexible and robust computational finance workflow that effectively links data
management, statistical modeling, technology, and decision-making.

5 It also needs to be consistent for all asset classes (rates, FX, equity, commodities, credit).

39

References




40

Bank of International Settlements (BIS), 2011, Basel III: A global regulatory framework for more
resilient banks and banking systems, Basel
Brigo, D. A. Dalessandro, M. Neugebauer, and F. Triki, 2007, A Stochastic Processes Toolkit for
Risk Management
Dodd Frank Act, 2010, Public Law 111 - 203 - Dodd-Frank Wall Street Reform and Consumer
Protection Act, U.S Government Printing Office
Office of the Comptroller of the Currency (OCC), 2011, Fed OCC guidelines handbook, U.S.
Department of Treasury
Zhu, S. H. and M. Pykhtin, 2007, A Guide to Modeling Counterparty Credit Risk, GARP Risk
Review, July/August 2007, available at SSRN: http://ssrn.com/abstract=1032522

High-level Debate

Aligning Marketing
and Finance with
Accepted Standards
for Valuing Brands
James Gregory Founder and CEO of CoreBrand
Michael Moore Professor in Residence of Accounting at the Loyola Marymount University

Abstract
With this paper we hope to create awareness and support
among the financial community for the need to have consistent brand measurement and metrics that tie investments
in corporate and product brands to financial value. There
are many obstacles to overcome before the financial community accepts such brand value measurements. Our initial
proposal suggests that brand value created (or lost) by the
investment in brands be reported in the Management Discussion and Analysis section of the financial reports. This will
promote a consistent methodology for valuing brands for internal management purposes and will provide to all external
parties full disclosure about the benefits and value created
(or lost) by investment in brands.

41

Introduction

The current landscape

There has been considerable effort by the Marketing Accountability Stan-

The implementation of a goal of reporting brand values is not without

dards Board (MASB) to raise awareness regarding financial reporting of

hurdles since a significant change in accounting practice and reporting

marketing intangibles. In many cases, these intangibles have grown to

will be necessary. Both the Financial Accounting Standards Board (FASB)

be of significant value, but receive little attention in internal financial and

and the International Accounting Standards Board (IASB) have been very

managerial accounting records, nor are they reported to external users

reluctant to depart from the current practice of deducting most advertis-

of financial statements. Yet, boards of directors and management are al-

ing and marketing expenses as ordinary operating expenses.

locating resources for marketing activities of the firm with little guidance
as to the values created by these expenditures.

For accounting (and tax) it is assumed that advertising costs incurred in


anticipation of future probable economic benefits are usually expensed

Investors and analysts attempt to value and compare firms and try to

currently because the benefit period is presumed to be short or the pe-

predict how assets owned by the firms will produce future income with

riods in which economic benefits might be received or the amount of

little knowledge of the value of marketing intangibles. MASB was cre-

economic benefit cannot be determined easily and objectively. As such,

ated by the Marketing Accountability Foundation as an independent pri-

internally developed (self-created) assets are not recognized as assets in

vate sector, self-governing body where marketing and finance align on

the accounting records. Intangible assets acquired in a business com-

measurement for reporting, forecasting, and improving financial returns

bination are measured and reported in the financial records under both

from buyers in markets. MASBs mission is to establish marketing mea-

FASB and IASB standards.

surement and accountability standards across industry and domain for


continuous improvement in financial performance and the guidance and

Marketing assets

education of business decision makers and users of performance and

With respect to branding and other marketing intangibles there are no-

financial information.

ticeable inconsistencies between the treatment of purchased intangibles


and internally developed intangibles. Additional characteristics should

MASB has several projects underway designed to raise awareness of the

also be noted. Internally developed intangibles are always carried on the

benefits and obstacles of formalizing the recognition of the brand as a

books at adjusted historical cost and are not written up to market. Like-

major marketing intangible.

wise, purchased intangibles are recorded at cost and are adjusted down
by amortization or impairment and never written up to market.

The Brand Investment/Valuation Model (BIV) Project has the objective


of providing the critical missing link between marketing and financial

The current debate about whether or not intangible assets such as the

communities by developing consistent, credible, and actionable brand

brand can be recorded on the balance sheet in situations where eco-

valuations through the establishment of generally accepted brand in-

nomic value was created as a result of prior expenditures shows a large

vestment and valuation standards, using metrics that are simple, trans-

gap between those who espouse such treatment and the policy makers

parent, relevant, and calibrated across categories and conditions and

at FASB and IASB. Recording assets at fair value using Generally Ac-

that reliably tie marketing actions to customer impact, to market out-

cepted Accounting Principles (GAAP) or International Accounting Stan-

comes, and to financial returns both short term and over time.

dards (IAS) is generally limited to marketable securities and write-downs


for asset impairments.

The Brand Investment/Valuation Marketing Communication Project has


a goal of effectively clarifying and communicating the operational impor-

It is felt by many marketing professionals that the balance sheet does

tance of the BIV Project to all constituencies.

not adequately reflect the fair value of marketing assets or in most cases
does not even reflect the existence of an asset. Most feel that such a

The Improving Financial Reporting (IFR) Project has as its goal to facili-

treatment falls short of information necessary for evaluation of areas such

tate partnering with the financial reporting and investment communities

as marketing effectiveness, investment and portfolio optimization, asset

for improving the accounting and/or reporting rules related to marketing

management, and benchmarking.

such that financial returns from corporations will be driven and measured
by buyer behavior in markets over time and to ensure that marketing is

Recording marketing assets on the balance sheet under mark-to-market

at the table when reporting of brand value is required for internally de-

accounting may not occur for several years. Development of a model for

veloped brands.

recording internally developed marketing assets that measures fair value


is critical to its acceptance by FASB and IAS. The fair value concept

42

focuses on the price that would be received upon the sale of an asset or

The Capco Institute Journal of Financial Transformation


Aligning Marketing and Finance with Accepted Standards for Valuing Brands

paid to transfer a liability. It is an exit price rather than a price that would

Level 3 inputs are unobservable inputs for an asset or liability.

be paid to acquire the asset or received to assume the liability, called an

Unobservable inputs are used to measure fair value to the extent

entry price. In other words, fair value is a market-based measurement,

observable inputs are not available. This allows for situations where

not an entity-specific measurement. A fair value measurement should be

there is little or no market activity for the asset or liability at the mea-

determined based on assumptions that market participants would use in

surement date. Unobservable inputs are developed based on the

pricing an asset or liability.

best information available in the circumstances. This might include


the reporting entitys own data. Unobservable inputs are intended to

Valuation issues

allow for situations in which there is little, if any, market activity for the

For both FASB and IAS, valuation techniques that are consistent with

asset or liability at the measurement date [FASB (2006), IFRS (2011)].

the market approach, income approach, and/or cost approach are used
to measure fair value. A market approach uses prices and other relevant

In addition to the general fair value models from FASB and IAS and specific

information generated by market transactions involving identical or com-

branding commercial valuation models such as Brand Finance, CoreBrand,

parable assets or liabilities. A valuation technique consistent with the

Interbrand, and Millward Brown, other models have been proposed, nota-

market approach might be one that uses market multiples derived from a

bly International Organization for Standardization (ISO), Monetary Brand

set of comparables. The income approach uses valuation techniques to

Valuation (ISO 10668), and International Valuation Standards which issued

convert future amounts to a discounted present value amount. The tech-

International Valuation Standards (IVS) -2011. This study sets standards for

niques include present value models, option-pricing models, lattice mod-

the valuation of a wide range of assets, including intangibles.

els, and the multi-period excess earnings method. The cost approach
is based on the amount that currently would be required to replace the

In order to persuade the FASB and the IAS that marketing assets should

service capacity of an asset (current replacement cost).

be reported at fair value in the financial statements, a valuation model must


be developed that has the characteristics consistent with fair value inputs

There are many inputs (assumptions) that market participants use in pric-

prescribed by FASB and IAS. MASB is currently working on validating such

ing an asset or liability. These include assumptions about risks inherent in

a fair value model. This project involves creation of general principles and

a valuation technique and/or the risks inherent in the inputs to the valuation

standards/methodologies for investing in and valuing brands. It will involve

technique. These inputs are classified as observable and unobservable.

empirical trials among three to five brands to serve as examples of ap-

Observable inputs are inputs that reflect market participant assumptions

plying the standards/methodology. The BIV methodology is based on an

based on market data obtained from sources independent of the reporting

income approach or cash flow approach that uses valuation techniques to

entity. Unobservable inputs are inputs that reflect the reporting entitys own

convert future amounts to a discounted present value amount. This model

assumptions about market participant assumptions that would be used

will have several levels of impact metrics: customer level, market level,

based on the best information available in the circumstances. There is a hi-

operating financial level, and non-operating financial level.

erarchy that prioritizes the inputs to valuation techniques used to measure


fair value. These are prioritized into three broad levels.

Fair value and financial statement


There are three approaches to the treatment of fair value write ups and

Level 1 inputs are quoted prices in active markets for identical assets

write downs on the financial statements. Using investments in debt and

or liabilities that the reporting entity has the ability to access at the

equity securities as an example, debt securities that a firm has the intent

measurement date.

and ability to hold to maturity are classified as held-to-maturity and re-

Level 2 inputs are inputs other than quoted prices included within

ported at amortized cost less impairment on the balance sheet (historical

Level 1 that are observable for the asset or liability, either directly or

cost approach). Debt and equity securities that are purchased principally

indirectly. Level 2 inputs include the following:

to sell in the near term are classified as trading securities and reported

Quoted prices for similar assets or liabilities in active markets.

at fair value on the balance sheet (mark-to-market approach). Unrealized

Quoted prices for identical or similar assets or liabilities in markets

gains and losses are included in earnings. Debt and equity securities not

that are not active.

classified as above are classified as available for sale securities and re-

Inputs other than quoted prices that are observable for the asset or

ported at fair value (mark-to-market approach) on the balance sheet. Un-

liability (for example, interest rates and yield curves observable at

realized gains and losses are reported as other comprehensive income,

commonly quoted intervals, volatilities, prepayment spreads, loss

which is reported in a separate component of shareholders equity.

severities, credit risks, and default rates.

Inputs that are derived principally from or corroborated by observ-

Problems abound even with fair value balance sheet measurements that

able market data by correlation or other means.

are currently acceptable methods of accounting. In a recent Wall Street

43

Journal article it was reported that the Public Companies Accounting

The simple inclusion of these notes, either separately or together, in the

Oversight Board (PCAOB) found 123 audit deficiencies of 250 audits in

MD&A section of the annual report will have a profound impact on the

2010 among clients of the big four accounting firms that were related

practice of marketing and not cause any disruption to current accounting

to fair-value estimates and asset impairments in 2010 [Chasan (2012)].

practice for external users. This single change will make marketing more

PCAOB questioned the assumptions and methodologies that went into

accountable and unite finance and marketing toward creating a common

some of the asset pricing models.

goal increasing enterprise value.

Marketing assets do not fit well in this current framework of fair value

Additional benefits include the consistent diagnostic evaluation of the

measurement for inclusion of such assets on the balance sheet and many

long-term financial health of the company. It will provide a dashboard

issues still need to be resolved by MASB and accounting policy makers.

measure by which the company management, investors, and employ-

MASB realizes that acceptance of fair value reporting of marketing assets

ees alike would be able to determine if the investments being made in

on the balance sheet may take a number of years.

marketing were wise and provided the company with an effective return.

A need for better reporting

Marketing activities can be highly efficient for value creation, but not

In the meantime the need exists for a consistent dashboard measure

when accounting standards obstruct reasonable requests. Valuation

for management and investors who need to know if a firm is creating

standards for intangible assets such as brands need to be recognized to

value through investments in advertising and other marketing activities.

finally get marketing and finance on the same page.

For reasons stated above, rather than entering the fair value of marketing
assets on the balance sheet, MASB is proposing that disclosure be made

What is in it for the CFO?

in the Managements Discussion and Analysis of Financial Condition and

It depends on whether individual CFOs define their role by providing ac-

Results of Operations (MD&A) section of the firms annual report. The

counting reports, or by providing financial leadership. Though intangible

MD&A seems to be a solid first step in encouraging development of dis-

assets have grown in value there is less understanding than ever of what

closure of marketing expenditures and results as seen through the eyes

drives that value. There is no doubt that accounting standards for intan-

of management.

gible assets will eventually be changing worldwide. CFOs who see their
role from the accounting perspective will wait until IASB standards for

Examples:

valuing intangibles are updated and distributed sometime in the distant

The following note is an example of a suggested MD&A disclosure for

future. But from the financial leadership perspective, CFOs will embrace

product brand values:

and encourage forward thinking ideas about what drives the value of intangible assets. Those CFOs who lead the charge will be providing better

As of June 30, 2012, the senior management of Gadget Corpora-

acceptance of company value and future value estimates by investors,

tion values the Widget Brand at U.S$2.9B, up 7% from year ago,

analysts, employees, and management. CFOs will be at the forefront of

and 20% over the past three years. We estimate this value using the

helping marketing and finance to work together toward common goals

methodology provided by ValuePack LLC,a MASB qualified Brand

with clear ROI measures in place.

Valuator.

Conclusion
The following note is an example of a proposed MD&A disclosure for

We believe the establishment of brand valuation standards will help com-

corporate brand values:

panies make better investment decisions, meet organic growth targets


more often, improve performance as measured by customer, market and

We, the senior management of Alpha Corporation, believe the value

financial outcomes, build strong brands more profitably and consistently,

of our corporate brand, as of December 31, 2011, is $26.4 billion, up

and will serve as the primary forward looking marketing KPI in Corporate

2.7% from a year ago, and down 3.8% over the past three years. We

Scorecards and in MD&A discussions.

estimate this brand value using the methodology provided by BrandTop, LLC,a MASB qualified Brand Valuator.

We seek to have CFOs, CPAs, and the world of finance join MASB to
achieve consistent, comparable, credible and actionable brand valua-

44

Both examples utilize continuous quantitative research study models

tions for both externally and internally developed corporate and product

based on how investments in the product or corporate brand impact fu-

brands. MASB is establishing generally accepted brand investment and

ture cash flows.

valuation standards using metrics that are simple, transparent, relevant,


and calibrated across categories, cultures, and conditions. The ultimate

The Capco Institute Journal of Financial Transformation


Aligning Marketing and Finance with Accepted Standards for Valuing Brands

goal is to tie reliably marketing actions to customer impact, market outcomes, and financial returns both short term and over time.
For further information please visit: http://www.themasb.org

References

Chasan, E., 2012, Asset Valuations Trip Up Audits, Wall Street Journal, 22 May
Financial Accounting Standards Board (FASB), 2006. Fair Value Measurements, Statement of
Financial Accounting Standards No. 157, available at: www.fasb.org

45

46

High-level Debate

CCP Regulations:
Posing Challenges
for Both CCPs and
Regulators
Jennifer Liu Principal Consultant, Capco

Abstract
Given the focus on central counterparty (CCP) clearing in
the post-financial crisis world, it is no surprise that clearing
houses today face more scrutiny from regulators and central banks than ever before. However, given the complicated
legal entity and organizational structure of CCPs, regulating
and providing oversight of clearing corporations can be challenging. This is because different subsidiaries tend to fall into
different jurisdictions and are subject to different regulatory
bodies. In addition, in light of the different products cleared,
CCPs can be subject to product-specific regulations. For
clearing houses, the complexity of complying with different
regulations across the world can be taxing. For regulators,
the intricate structure of global clearing houses poses challenges when applying appropriate oversight.

47

Introduction

Background

Importance of CCPs in todays regulatory environment

CCP history and evolution

In the wake of the financial crisis and in response to the G20 commit-

CCPs have long been used by derivatives exchanges and a few securities

ment to provide global financial stability, regulators around the world

exchanges. In recent years, they have been introduced into more securi-

are increasingly looking to central counterparty (CCP) clearing houses

ties markets, including cash markets and OTC fixed income markets, be-

as a way to mitigate counterparty risk in the market. For example, the

coming an integral part of the financial market infrastructure. For example,

key commitments to financial market reform stated: All standardized

repurchase agreements, also known as repos and an important financial

OTC derivatives contracts should be traded on exchanges or electronic

instrument used to provide funding for dealers and banks, can now be

trading platforms, where appropriate, and cleared through central coun-

cleared through several clearing houses in Europe, the US, and Canada.

terparties (CCPs) by end 2012 at the latest. OTC derivatives contracts


should be reported to Trade Repositories. Non-centrally cleared contracts

The history of clearing houses, providers of central counterparty (CCP)

should be subject to higher capital requirements. Proposals in both the

clearing services, can be traced back to more than 100 years ago when

US Dodd-Frank Act and the European Market Infrastructure Regulation

earlier clearing houses cleared the commodities derivatives in local mar-

(EMIR) stress the importance of having over-the-counter (OTC) derivative

kets. For example, London Clearing House (LCH) was established to

transactions cleared through CCPs.

clear commodities contracts in London in 1888 while ICE Clearing US


(InterContinentalExchange) was founded in 1915 as the New York Cotton

Central counterparty clearing refers to the process by which financial

Exchange Clearing Association.

transactions are cleared by a centralized counterparty. Through a process called novation, the CCP becomes the seller to the buyer and the

The expansion of clearing houses is a story of mergers and acquisitions.

buyer to the seller. A clearing house usually provides the CCP service

LCH was merged in 2003 with Clearnet SA, which was formed in 1969

and manages the counterparty default risk by leveraging its built-in risk

to clear Paris commodity contracts, to become LCH.Clearnet Group. It

management mechanisms, such as margins or default funds contributed

now clears exchange traded and over-the-counter (OTC) derivatives and

by its clearing members.


Initially devised as a mechanism to reduce transaction costs by calculat-

Clearing houses

Location

Sample cleared products

ing members net obligations to post margins and settle contracts, the

LCH.Clearnet

UK
Europe
US

IRS (Interest Rate Swaps)


ccCFDs
Commodities
CDS (Credit Default Swaps)
Derivatives
Energy
Equities
Fixed Income
Freight
Foreign Exchange

Eurex Clearing

Europe
UK

Derivatives
Equities
Bonds
Repo
Secured funding
Securities financing
Energy transactions

CME Clearing

US
Europe (through
CME Europe)

IRS (Interest Rate Swaps)


CDS (Credit Default Swaps)
FX
Zero coupon swap
Overnight index swap (OIS)
Basis swap
Forward rate agreement
Commodities

ICE Clearing

US
Europe
Canada

Futures (including Agricultural, FX,


financial index, energy)
CDS

role of the CCP has expanded over time. Since the financial crisis, the
importance of CCPs in providing financial stability through their risk management arrangements cannot be overstated.

Central clearing is an essential part of regulations post crisis


CCPs are referenced in almost all the major financial system reforms and
regulatory proposals post crisis, including Dodd-Frank, EMIR, BASEL III
Shadow Banking, and others. Examples include:

Dodd-Frank: Mandatory central clearing of standardized OTC


derivative contracts

European Market Infrastructure Regulation (EMIR): Central clearing for


certain classes of OTC derivatives are mandated (also supported in
Markets in Financial Instruments Directive 2 (MiFid 2))

BASEL III: Strong incentives for market participants to go through


central clearing including:

Higher capital charges on bilaterally cleared trades

Preferential treatment of centrally cleared trades

Encourages movement of OTC derivatives to CCPs, thus reducing


CVA capital charges

FSB (Financial Stability Board) Shadow Banking Regulation: Evaluating


the establishment or wider-use of central clearing where appropriate

48

for securities lending and repo

Source: LCH.Clearnet, Eurex Clearing, CME Group, ICE Clearing

Table 1 Examples of CCP location and clearing product offerings

The Capco Institute Journal of Financial Transformation


CCP Regulations: Posing Challenges for Both CCPs and Regulators

other securities worldwide. ICE Group acquired Clearing Corporation in

Holding company

2009 to provide the risk management framework, operational processing, and clearing infrastructure for ICE Clear Credit and ICE Clear Europe.
CDCC (Canadian Derivatives Clearing Corporation) is the product of a
merger between Montral and Toronto options clearing houses in 1977

LCH.Clearnet Group Limited


(Incorporated in UK)
Regulated as a compagnie financire by the Autorit de Contrle Prudentiel (ACP) (France)

and recently launched a fixed-income CCP service in addition to its derivatives clearing business.

Major global CCPs are diverse in locations and product


offerings
Major global CCPs such as LCH.Clearnet, CME, ICE, and Eurex have
multiple locations across the continents and have a variety of product
offerings (Table 1).

Operating entities

LCH.Clearnet LLC
Location: New York
Organized in Delaware
Regulated as a
Derivatives Clearing
Organization (DCO)
by the US Commodity
Futures Trading
Commission (CFTC).

How CCPs are regulated


Given the complicated legal entity and organizational structure of CCPs,

LCH.Clearnet Limited
Location: London
Incorporated in UK

LCH.Clearnet SA
Location: Paris
Incorporated in France

Regulated as a
Recognized Clearing
House by the Bank of
England, as a Derivatives
Clearing Organization
(DCO) by the US
Commodity Futures
Trading Commission
(CFTC).

Regulated as a Credit
Institution and Clearing
House by a regulatory
college consisting of the
market regulators and
central banks from the
jurisdictions of France,
Netherlands, Belgium and
Portugal.
Regulated as a
Recognized Overseas
Clearing House by the
Bank of England and by
other market regulators
and central banks in
jurisdictions in which
business is carried out.
Pending Derivatives
Clearing Organization
(DCO) application with
the CFTC.

Regulated by other
market regulators
and central banks in
jurisdictions in which
business is carried out.
Payment systems are
overseen by the Bank of
England.

regulating and providing oversight of clearing corporations can be challenging as different subsidiaries tend to fall into different jurisdictions and
are subject to different regulatory bodies. In addition, given the different
products cleared, CCPs can be subject to product-specific regulations.

Regulation by location
From a regional perspective, in Europe clearing houses with entities located or business conducted in the UK are regulated by the Financial Services

Luxembourg
subsidiary

Authority (FSA), with payment systems overseen by the Bank of England.


Examples include LCH.Clearnet Ltd, CME Europe, and ICE Europe.
Entities located or incorporated in other European jurisdictions are also

Source: LCH.Clearnet

Figure 1 LCH.Clearnet organizational structure

subject to their local regulators. For example, Eurex Clearing is a company incorporated in Germany and licensed as a credit institution under

This means CCPs that have businesses in multiple jurisdictions and

supervision of the Bundesanstalt fr Finanzdienstleistungsaufsicht (Ba-

multiple product offerings are subject to different regulatory and legal

Fin) pursuant to the Banking Act (Gesetz fr das Kreditwesen).

requirements. Take LCH.Clearnet for example. The holding company is


regulated as a compagnie financire by the Autorit de Contrle Pruden-

In the US, clearing houses such as CME Clearing and FICC are registered

tiel (ACP) (France). Its operating entities are regulated by CFTC in the US,

and regulated by the Securities and Exchange Commission (SEC).

Bank of England in the UK, and a regulatory college consisting of the


market regulators and central banks from various European jurisdictions,

Regulation by products

depending on where business is conducted by the entities. See Figure 1

From a product perspective, CCPs can be subject to certain regulation

for illustration.

depending on their product offerings. For example, the US Commodseeks to provide clearing services with respect to futures contracts, op-

New regulatory requirements post crisis and their


implications for CCPs

tions on futures contracts, or swaps to register with the CFTC as a de-

After the crisis, regulators in the world worked in tandem to review the

rivatives clearing organization (DCO) before it can begin providing such

existing financial infrastructure and drafted new regulations that aim to in-

services. As a mandate to implement the Dodd-Frank Act, this require-

crease transparency and stability of the global financial systems. In 2012,

ment not only affects clearing houses in the US but also in the rest of the

CPSS (Committee on Payment and Settlement Systems) and IOSCO (In-

world. Most clearing houses are registered or applying to be registered

ternational Organization of Securities Commissions) published the Prin-

with the CFTC now, including CME Clearing US, Eurex Clearing, LCH.

ciples for Financial Market Infrastructures, including updated principles

Clearnet, and ICE Clear (US, Europe, and Credit).

for CCPs such as higher financial resources and collateral requirements,

ity Futures Trading Commission (CFTC) requires any clearing house that

49

more robust and frequent stress tests, and processes for orderly resolution

EMIR, CCPs are required to publicly disclose information such as

of Clearing Member positions in the event of default. These principles are

fees, risk model data, aggregate transaction volumes, rules and pro-

followed and reflected in the Dodd-Frank and EMIR regulations.

cedures, and organizational structures. CCPs also need to ensure that


clearing members, clients, and regulators can access information as

These rules and directives enhance existing requirements as well as new

required. Under Dodd-Frank, there are specific reports required by the

requirements for CCPs including:

CFTC daily, monthly, quarterly and ad hoc.

Segregation and portability Segregation of client asset and porta-

Impact for CCPs: The increasing amount of information and data re-

bility is enforced to safeguard the assets of clearing members clients

quired by the regulators, as well as the frequency of reporting, may

in the case of clearing member default. For example, in Europe, under

require the CCP to build additional systems to meet the requests. The

EMIR, CCPs must offer both omnibus and individually segregated

required accessibility may also demand additional development effort

client account structures for both assets and positions across all
asset classes. Portability must also be offered and possible for each

in the CCP technology team.

Business continuity EMIR and Dodd-Frank also addressed busi-

account structure.

ness continuity requirements that CCPs need to meet. These state

Impact for CCPs: Technology development may be needed to ensure

that recovery of a CCPs critical functions must be possible within two

required account structure. Portability process needs to be set up or

hours including extreme circumstances. Also, a back-up data center

enforced.

should be maintained at a remote location.

Reporting to trade repositories and record-keeping Under

Impact for CCPs: The heightened standard for business continuity

Dodd-Frank and EMIR, swaps or transactions of all asset classes are

management (BCM) means that CCPs may need to have more re-

required to be reported to trade/data repositories or market authori-

sources and technology support in place to meet the requirement.

ties (e.g., European Securities and Markets Authority (ESMA)). This


need to report them on behalf of their clearing members.

Challenges for both CCPs and regulators in the


new regulatory environment

Impact for CCPs: Technology development needed to comply with the

In the new regulatory environment, CCPs are facing challenges in their

reporting requirements. Different requirements from different trade/

efforts to comply with regulators requirements. These challenges arise

data depositories can pose a challenge for CCPs to meet all the re-

from the need for resources and technology development, as well as the

quirements in time.

availability of regulatory information.

means CCPs need to report the transactions as required and may also

Default fund and waterfall In Europe, EMIR has established new


standards for a CCPs stress testing framework, which is used to

Technology development The impact of the regulatory mandates

calculate the minimum size of its default fund. Also, CCPs are now

means that technology will play a critical role for CCPs aiming to meet

required to hold a minimum amount of dedicated own resources in

the requirements. However, building a technology infrastructure to

the risk waterfall.


Impact for CCPs: CCPs will have more skin in the game in any sce-

meet these stipulations can be a lengthy and costly exercise.

Resources To meet the regulatory requirements, CCPs need staff

nario where a clearing member defaults. This requires CCPs to moni-

with knowledge of regulatory compliance and expertise in the clearing

tor closely the risk of their clearing members.

business. However, given the rapidly evolving regulatory environment


and the relatively new focus on central clearing, such subject matter

Some of the existing rules are more stringent and enhanced:

Capital requirements CCPs now have more stringent capital

experts are often a scarce commodity in the market.

Information availability from regulators Last but not least, CCPs

requirements. In Europe, under EMIR, CCPs must hold sufficient capi-

must have clear, adequate information on the interpretation of the

tal to wind down or restructure their activities, with a minimum of six

rules and the implementation effort required. However, in many cases

months operational expenses to cover various risks. In the US, under

the rules are still evolving so timelines, priorities, and scope will be

Dodd-Frank, CCPs are required to hold enough capital to operate for

ambiguous for the foreseeable future.

one year, including six months in highly liquid capital.

50

Impact for CCPs: The capital required of CCPs comes at a cost. While

On the other side of the coin, regulators are facing their own challenges

CCPs try to meet the capital requirement, they also need to under-

drafting and effectively enforcing the new requirements concerning CCPs:

stand the economic consequences and the balance between compli-

Complication of CCP organization structures CCPs are orga-

ance and business growth.

nized differently and may have different ownership structures. For

Transparency and Regulatory Reporting Regulators require CCPs

example, some clearing houses structure their business as a matrix of

to be more transparent and increase regulatory reporting. Under

geographies, products and markets (e.g., LCH.Clearnet, Eurex), while

The Capco Institute Journal of Financial Transformation


CCP Regulations: Posing Challenges for Both CCPs and Regulators

others have different regional subsidiaries and leverage the infra-

Creation of new CCPs In addition, new CCPs might be created

structure of other subsidiaries (e.g., CME Europe). Also, some CCPs

in response to market demand and regulatory reform. For example,

are organized by geography but also have subsidiaries focused on

Canada launched a Fixed Income CCP in 2012. Hong Kong Exchange

specific products (e.g., ICE Clearing). Regulatory authorities need to

and Clearing (HKEx) planned to launch OTC derivatives CCP to meet

understand the structure of CCPs and all their subsidiaries to ensure

Hong Kong Monetary Authoritys mandate that all standardised OTC

all groups within CCPs are properly regulated.

derivatives transactions be cleared at central counterparty (CCP)

Balance between safeguarding and CCP economic viability

clearing facilities.

Given the significance of CCPs in the market, especially when the


market is under duress, the risk management and operations of

These trends inevitably pose challenges for regulators as they endeavor

CCPs need to be strictly regulated and closely monitored. However,

to apply proper oversight and enforcement of the rules. For example,

complying with the rules may have a significant impact on CCPs

as participation in CCPs increases, especially from buy-side institutions,

capital costs and operating costs. This includes the cost of technol-

regulators from buy-side as well as sell-side need to be involved when

ogy development and additional support staff. CCPs may even be

reviewing regulatory requirements for the CCPs.

challenged to remain economically viable. It can be argued that some


when CCP clearing costs deter market participants, the volume of

Considerations for CCPs and regulators in the new


world order

non-mandated products clearing through the CCP may decrease, and

Consideration for CCPs

even the market growth of products that are mandated to be centrally

To meet the challenges in the post-crisis regulatory environment and stay

cleared may slow down.

ahead of the game, CCPs need to look internally and externally for op-

CCPs are able to pass the cost to their clearing members. However,

portunities to grow their business and optimize operations.

Future challenges from market growth


As CCPs try to grow their business through business development and

Internally, CCPs can evaluate the following:

operational efficiency optimization, the following trends are observed in

the market:

compliance teams by subsidiaries or locations. There is an opportu-

Increasing participation More market players are participating

nity to increase collaboration and communication among the teams,

in the CCPs. This includes buy-side clients, not just bank/dealers.

if they are not already globally integrated, and create synergies in

Different participation models are being developed to accommodate


different participants.

knowledge while sharing experience.

to meet different regulatory requests, CCPs need to have a stra-

products and currencies to their clients, especially in the OTC deriva-

tegic roadmap and carefully plan out and prioritize the technology

tives space given the demand from banks to meet the central clearing

infrastructure build. The technology group may also want to choose

requirements. In addition, the broader range of products that a clear-

between an in-house build versus third party vendors who may

can achieve through cross-margining.

already have the expertise and a strong platform in place.

Resource investment As a long term strategy, CCPs should con-

Geographic expansions Clearing houses have been expanding

sider investing in resources with both regulatory and CCP expertise,

their footprints into different regions, primarily between the US and

or investing in educating staff and helping them become experts. In

Europe. For example, LCH.Clearnet opened its US office in 2009 and

the interim, CCPs may also consider forming alliances with consulting

CME Clearing Europe was launched in 2011. It is easy to predict that


new territories, such as Australia, Asia, and South America, will be

Technology build With all the technology development required

Product extensions CCPs are striving to provide a broader suite of

ing house can offer to its clients, the more capital efficiency its clients

Compliance team structure Currently most CCPs have separate

firms that have strong expertise in this area.

Operating model CCPs can use the regulatory requirements as

destinations for clearing houses searching for new business oppor-

an opportunity to revisit their operating model and identify any inef-

tunities.

ficiency in processes, application management, and allocation of

Continuing merger and acquisitions Since beginning of the year,

resources. They will then need to formulate strategies that help opti-

there have been further market infrastructure mergers and acquisi-

mize their operations.

tions, and we can expect to see even more activities in this area. For
example, ICE and NYSE announced their merger in December 2012.

Externally, CCPs need to maintain a close dialogue with regulators to

Also, EuroCCP, the European unit of the DTCC (Depository Trust &

keep abreast of the latest regulatory requirements. They also need to

Clearing Corp.) and EMCF (European Multilateral Clearing Facility)

be up to date with industry trends and be able to leverage industry best

plan to merge and form the biggest pan-European clearinghouse.

practices.

51

Current state
Jurisdiction A
Authority
X

Authority
Y

Jurisdiction B
Authority
Z

Authority
X

Authority
Y

Subsidiary 1

Jurisdiction C
Authority
Z

Authority
X

Subsidiary 2

Authority
Y

Authority
Z

Subsidiary 3

CCP
Regulation within jurisdiction
Regulation across jurisdiction

Future state
Jurisdiction A
Authority
X

Authority
Y

Jurisdiction B
Authority
Z

Authority
X

Subsidiary 1

Authority
Y

Jurisdiction C
Authority
Z

Subsidiary 2

Authority
X

Authority
Y

Authority
Z

Subsidiary 3

CCP

Collaboration/communication among regulatory authorities


Collaboration/communication among CCP subsidiaries
Feedback loop between CCP and regulatory authorities
Figure 2 Current and proposed future state for regulators and CCPs

Considerations for regulators

is when products currently cleared by CCPs are assessed to deter-

Given the complexities involved in rule making and enforcement related

mine whether they should be subject to the clearing obligation. Going

to CCPs, regulators should continue working with CCPs, increase col-

forward, dynamic feedback between regulators and CCPs, as well as

laboration among regulatory bodies, and have a holistic understanding of


the potential impact of CCP regulations on the whole market.

industry-wide open forums, will facilitate dialogues and collaboration.

Collaborations among different regulatory bodies Central banks,


securities regulators, supervisors and other relevant authorities

52

Collaborations with CCPs By working with CCPs, regulators can

should increase collaboration to ensure comprehensive coverage of

develop a good understanding of their organizational structures and

CCPs, not only within the same jurisdictions but cross-border and

the implementation efforts and cost implications. Currently, both top-

globally. For example, the Dodd-Frank Act provides that the CFTC will

down and bottom-up approaches are adopted when rules are pro-

regulate swaps, SEC will regulate security-based swaps, and both

posed and drawn up. An example of the top-down approach is when

the CFTC and SEC will regulate mixed swaps. This requirement

authorities call for clearing to be introduced for products where no

inevitably requires close collaboration between CFTC and SEC. By

CCP currently provides services. An example of bottom-up approach

aligning on their approaches and priorities, regulators can help CCPs

The Capco Institute Journal of Financial Transformation


CCP Regulations: Posing Challenges for Both CCPs and Regulators

focus on key priorities and provide clarification in the case where a


single CCP is monitored by multiple regulators. Figure 2 illustrates an

References

example of a future state where regulatory authorities will collaborate


with each other and with CCPs.

Understand the broader impact in the whole market When considering new rules, regulators need to carefully evaluate the potential

impact on the markets and avoid unintended consequences. As

mentioned earlier, the increasing cost of central clearing may curb the

growth of certain market segments. This could also have an impact

on liquidity the financial markets. For example, it is estimated that the


increase in collateral required by CCPs for initial margin and default
fund contributions to clear standardized OTC derivatives trades can
be up to US$2 trillion according to TABB Forum. This may create
additional liquidity strain in the market, as cash and highly liquid collateral required by CCPs will be held in the custody accounts.

Summary
Regulation of CCPs is a critical task in todays global market. CCPs have

CFTC Office of Public Affairs, 2012, Q & A Proposed Rules and Interpretive Guidance Further
Defining Swap, Security-Based Swap, and Security-Based Swap Agreement; Regarding
Mixed Swaps; and, Governing Books and Records for Security Based Swap Agreements,
Commodity Futures Trading Commission
Commodity Futures Trading Commission (CFTC), 2011, Derivatives Clearing Organization
General Provisions and Core Principles, Federal Register, 76, 216
CPSS-IOSCO, 2012, Principles for Financial Market Infrastructures, Bank of International
Settlement and International Organization of Securities Commissions
DoddFrank Wall Street Reform and Consumer Protection Act, 2010
The Eurofi, 2012, Improving Collateral Management Efficiency in the Context of OTC Derivative
Reforms, The Eurofi Financial Forum, September
The European Commission, 2013, Regulatory Technical Standards on Capital Requirements
for Central Counterparties, Official Journal of the European Union, February
The European Commission, 2013, Regulatory Technical Standards on Requirements for
Central Counterparties, Official Journal of the European Union, February
The European Commission, 2013, Implementing Technical Standards on Requirements for
Central Counterparties, Official Journal of the European Union, February
Financial Stability Board (FSB), 2012, Strengthening Oversight and Regulation of Shadow
Banking, An Integrated Overview of Policy Recommendations, Consultative Document

IntercontinentalExchange (ICE), 2012, IntercontinentalExchange to Acquire NYSE Euronext For


$33.12 Per Share in Stock and Cash, Creating Premier Global Market Operator, Press Release,
20 December
LCH.Clearnet, 2013, EMIR Dodd-Frank Q&A, Regulatory Update, February 2013
Sukumar, N., 2013, EuroCCP-EMCF to Merge, Forming Pan-European Clearinghouse,
Bloomberg, March 14

become an important component of the global financial market infrastructure. Unless properly managed and well capitalized, the CCPs can
potentially pose significant systemic risk to the market, as they take on
additional concentrated risk. There have been heated discussions about
the vulnerability of CCPs in distressed market situations. Some even
start to apply the term too big to fail to CCPs. More standardization
and transparency of CCPs can therefore be anticipated as the market
evolves. In the meantime, regulators need to have a holistic view of the
market infrastructure to effectively develop and enforce regulations that
ensures the safety of the market while encouraging the healthy growth of
market. Like the old saying Rome was not built in a day, we can expect
this to be a long, albeit exciting, journey.

53

54

High-level Debate

The US Home
Foreclosures and
Federal Funds Rate:
Is the Channel
Effect Muted?
Gke Soydemir College of Business Administration, California State University, Stanislaus
Andres Bello College of Business Administration, University of Texas-Pan American
Tzu-Man Huang College of Business Administration, California State University, Stanislaus

Abstract
We provide empirical evidence on the postulation that the
channel effect in the US is muted. The estimation results
from a multivariate model reveal that the movements in the
federal funds rate do not appear to have the intended timely
and pronounced impact on foreclosures starts and purchase
prices. The response of the home mortgage interest rates
to the federal funds rate however, is significant and positive
but the effect is not sustained. These findings are consistent
with the view the channel effect in the US is relatively muted.

55

Introduction

Theoretical background

In the summer of 2007, the US stock market crashed and brought with it

One of the fundamental differences between the various schools of

the housing market meltdown. The meltdown spread beyond the hous-

macroeconomic analysis is the representation of the supply side of the

ing market and became global resulting in a systemic crisis in 2008.1 The

economy. The nature and slope of the aggregate supply schedule can

Federal Reserve responded to avert the crisis and minimize its impact in

have an effect on the self-equilibrating or full-employment level of output

the US by initially utilizing traditional tools such as the federal funds rate.

properties of the macroeconomic system and whether or not changes in

Changes in the federal funds rate did not have an immediate impact on

aggregate demand plays any role in determining the equilibrium level of

the housing market.

output and prices [Stevenson et al. (1988)].

Alan Greenspan noted in the recent past that the Federal Reserve be-

For example, the Keynesian theory [Keynes (1936)] posits that under the

came acutely aware of the disconnect between monetary policy and

assumption of rigid prices, there is a case for intervention on the part of

mortgage rates when the latter failed to respond as expected to the Fed

policymakers by way of aggregate demand management. In other words,

tightening in mid-2004. Moreover, the data show that home mortgage

beginning from an under-employment equilibrium, fiscal and monetary

rates had become gradually decoupled from monetary policy even ear-

policy can be targeted to affect aggregate demand in a way that secures

lier in the wake of the emergence, beginning around the turn of this

full-employment level of output provided that the aggregate supply curve

century, of a well arbitraged global market for long-term debt instru-

has the normal upward sloping shape.4

ments.

The basic textbook Classical economics asserts that prices are comFurther, Rubio (2008) in investigating the postulated relationship between

pletely flexible (both in the short run and long run) and with a relative-

fixed and variable rate mortgages, business cycles, and monetary policy,

ly vertical aggregate supply curve, any changes in aggregate demand

quotes Ben Bernanke: [...] the structure of mortgage contracts may mat-

schedule brought about by fiscal and/or monetary policy can, at equi-

ter for consumption behavior. In countries like the United Kingdom, for

librium, only have an effect on the price level with no significant impact

example, where most mortgages have adjustable rates, the changes in

on the level of output. Thus in the Classical model, the levels of output

short-term interest rates have an almost immediate effect on household

and employment are determined solely by supply factors, such as the

cash flows. [...] In an economy where most mortgages carry fixed rates,

amount of output the firms choose to produce. Another general feature

such as the United States, that channel of effect may be more muted. I do

of Classical economics is the self-adjusting tendencies of the economy.

not think we know at this point whether, in the case of households, these

Government policies to ensure an adequate demand for output were

effects are quantitatively significant at the aggregate. Certainly, these is-

among those state actions considered by the Classical economists to be

sues seem worthy of further study.

unnecessary and even harmful [Froyen (1996)].

Utilizing two decades of data, the aim of this paper is to conduct empirical

Fiscal policy implications of the Classical model on the level of output are

tests on the postulation that the federal funds rates effect on US home

similar. Higher government spending increases the demand for loanable

foreclosures is relatively muted. Consequently, we make the following

funds. The resulting rise in the federal funds rate causes a decline in invest-

unique contributions to the existing literature. First we test whether there

ment and consumption but an increase in saving. However, the decline

is a causal relationship between the federal funds rate and foreclosure

in investment and consumption just balances the increase in government

activity and thus shed some light on an issue that has yet to be studied

spending, thus causing the crowding out effect to be complete. In Keynes-

extensively. We investigate whether the results differ in the US from one

ian economics, the crowding out effect is partial, and in the case of the

region to another. We utilize a multivariate model that takes into account

horizontal aggregate supply curve, the crowding out effect is zero.

information contained in the variance covariance matrix and do a better


job in capturing the data generating process than univariate models.

The production in the housing market constitutes part of the aggregate


level of output and changes in the federal funds rate reflect the monetary

The estimation results from a multivariate model indicate that the recent
movements in the federal funds rate do not appear to have intended
timely and pronounced impact on foreclosures. Further, the housing prices fail to respond to decreases in the federal funds rates. The findings are
consistent with the view that the correction in the housing market is likely
to continue as a policy independent process until all excess inventories
56

are fully depleted.

1 Financial Times, 2008, Credit turmoil flagged up as risk to global outlook, January 10
2 Greenspan, A., 2009, The Fed Did Not Cause the Housing Bubble, Wall Street Journal,
Opinion Column, March 11
3 Bernanke, B., 2007, The Financial Accelerator and the Credit Channel, Remarks at a
Conference at the Federal Reserve Bank of Atlanta.
4 In the extreme case of a horizontal aggregate supply curve, the price level remains constant
and aggregate demand management becomes most effective in securing full-employment.

The Capco Institute Journal of Financial Transformation


The US Home Foreclosures and Federal Funds Rate: Is the Channel Effect Muted?

policy stance of the Federal Reserve. Indeed the extent to which the fed-

more sellers in the market than buyers. As inventories of unsold houses

eral funds rate impacts the US home foreclosures may be questioned in

rise, creating excess supply, foreclosures rise. Naturally, this process

the context of the two schools of macroeconomic analysis.

would continue until all excess inventories are depleted.

In a simplistic closed economy Keynesian setting, income (Y) in an econ-

The remainder of this study is organized as follows. Section three de-

omy equals consumption expenditure (C), investment expenditure (I) and

scribes the data and econometric methodology. Section four reports em-

government expenditure (G);

pirical findings and section five concludes.

Y=C+I+G

(1)

Data and econometric methodology


Households can allocate this income in three different ways in the form of

The sample interval spans the third quarter of 1980 through the fourth

consumption, savings (S), or taxes (T);

quarter of 2012. All data are obtained in quarterly frequency from

Y = C + S + T

(2)

Datastream. The Fed policy proxy variable used in this study is the effective federal funds rate (FEDFUND). Foreclosure starts (FORCST) are for

We can assume that foreclosures (F) is related to savings by the following

all residential mortgage loans, as are the home purchase prices (PRICE).

behavioral equation in that higher foreclosures are negatively related to

Home mortgage rates are for the 30 year fixed rates (HOMIRATE). Re-

savings;

gional variables considered in the study are the foreclosure starts from

S = 0 - 1F 

(3)

the regions of the West (WEST), South (SOUTH), Northeast (NORTHEAST), and North Central (NORTHCENTRAL). Price series were not avail-

Investment expenditures have the usual negative relation to interest rates

able at the regional level from Datastream.

(r) expressed as;


I = 0 - 1r(4)

Figure 1 plots the foreclosure starts from the third quarter of 1980 to the
fourth quarter of 2012. It appears that the jump in foreclosures starts in

By substitution and solving for F, the reduced form can be expressed as;

the fourth quarter of 2006 with a spike from .47 in the third quarter to

F = (1/ 1)(0 - 0 + T G + 1r)

.57 in the fourth quarter, and then exhibiting an exponential rise. A visual
inspection of the graph shows that there may be evidence of a slight

where an increase in G would lead to a decrease in F by a proportion

leveling-off effect towards the latter part of the sample.

of (1/ 1) and an increase in T would lead an increase in F by the same


amount. An increase in interest rates would lead to an increase in F by 1/

Figure 2 plots foreclosures against the federal funds rate normalized by

1. When the null hypothesis holds, (1/ 1) = 0, interest rates (in this case

10 to make comparisons between the two series easier. It appears that

the federal funds rate) would be muted or have no effect on foreclosures

until the third quarter of 2006 both series long-run trend appears to be

whereas when the alternative hypothesis, (1/ 1) 0, the Keynesian pre-

either somewhat flat or slightly upward for foreclosure starts and down-

diction of the interest rates impact on foreclosures would hold.

ward for the Federal Funds rate consistent with either no presence of an
association between the two or a weak negative association. However,

Further, apart from the two schools of thought, there is some evidence in

toward the latter part of the sample, after the third quarter of 2006, the

the finance literature to suggest the overvaluation of the housing market

series appear to be moving clearly in the opposite direction of each other,

can be due to price dynamics rather than overreaction to fundamentals

perhaps suggesting a well-defined inverse correlation. Indeed a correla-

[Fraser et al. (2008)], a result in favor of the view that much of the price

tion coefficient check between the series produces a negative correlation

dynamics is an independently driven process.

coefficient of -.93 in this sub-sample period and -.58 in the entire sample. These findings are consistent with the view that during much of the

Foreclosures may also be linked indirectly to momentum investor behav-

sample there is no association or slight negative association between the

ior [Shiller (1984), Kyle (1985), DeLong et al. (1990), Daniel et al. (1998),

series. In the latter part it also suggests a negative association, contrary

Barberis et al. (1998), Hong and Stein (1999), Lui et al. (1999)]. Driven by

to the expectation of a positive association of a lowering of the funds rate

price alone, momentum occurs when agents buy after price increases

and falling of the foreclosure rates. These findings are also consistent for

and sell after price decreases. In this scenario agents expect a price rise

the variables in the form of logarithmic first differences which would cap-

or fall to continue. Thus in a down market price falls, quantity demanded

ture any leveling off effect or decrease in the rate of increase. To measure

decreases, buyers get discouraged. This expectation of falling prices

causality, a VAR model is used in the next stage.

causes demand to fall even further. An indirect result of this is the impact
on foreclosures. Continuing decrease in demand means there are now far

The vector autoregression (VAR) model employed in this study is useful

57

Number of
Cointegrating
equations

Value
1,6
1,4

None

1,2
1
0,8

Eigenvalue

Trace

P-value**

Maximum
Eigenvalue

P-value**

0.254

44.951

0.091

26.335

0.072

At most 1

0.111

18.616

0.521

10.598

0.687

At most 2

0.082

8.018

0.464

7.724

0.408

At most 3

0.003

0.294

0.588

0.294

0.588

0,6

* Series used FEDFUND, HOMIRATE, PRICE, FORCST


** MacKinnon-Haug-Michelis (1999) p-value estimates

0,4
0,2

Table 1 Johansen Cointegration Test*


Q3 1992
Q1 1993
Q3 1993
Q1 1994
Q3 1994
Q1 1995
Q3 1995
Q1 1996
Q3 1996
Q1 1997
Q3 1997
Q1 1998
Q3 1998
Q1 1999
Q3 1999
Q1 2000
Q3 2000
Q1 2001
Q3 2001
Q1 2002
Q3 2002
Q1 2003
Q3 2003
Q1 2004
Q3 2004
Q1 2005
Q3 2005
Q1 2006
Q3 2006
Q1 2007
Q3 2007
Q1 2008
Q3 2008
Q1 2009
Q3 2009
Q1 2010
Q3 2010
Q1 2011
Q3 2011
Q1 2012

Quarters

Figure 1 US residential market: All foreclosures started

Unit root tests are employed to analyze the time series properties of the
data to avoid the possibility of finding spurious relationships. In particular, the Augmented Dickey-Fuller (ADF) test [Dickey and Fuller (1979,
1981), Enders (1995)] fails to reject the null hypothesis of non-stationarity
in logarithmic levels for most of the series, but we reject the null for the

Value
1,6

log returns for all the variables. Given that most of the series are non-

1,4

stationary in levels, differences of the series are taken. The AIC and SBC

1,2

criteria suggest that the appropriate number of lags to be used is two and

11 respectively. However, given the limited number of observations in the

0,8

sample and to avoid any issues with over parameterization and thus the

0,6

degrees of freedom, a decision was made to limit the number of lags to

0,4

one and compare results from one to two and 11 lags.

Q3 1992
Q1 1993
Q3 1993
Q1 1994
Q3 1994
Q1 1995
Q3 1995
Q1 1996
Q3 1996
Q1 1997
Q3 1997
Q1 1998
Q3 1998
Q1 1999
Q3 1999
Q1 2000
Q3 2000
Q1 2001
Q3 2001
Q1 2002
Q3 2002
Q1 2003
Q3 2003
Q1 2004
Q3 2004
Q1 2005
Q3 2005
Q1 2006
Q3 2006
Q1 2007
Q3 2007
Q1 2008
Q3 2008
Q1 2009
Q3 2009
Q1 2010
Q3 2010
Q1 2011
Q3 2011
Q1 2012

0,2

Foreclosure starts

Federal Funds Target Rate/10

Figure 2 Foreclosure starts and the Federal Funds Rate/10

Table 1 reports results from Johansen (1988) cointegration tests. The


trace as well as the maximum eigenvalue statistics suggests that during the sample period, there is no evidence of cointegration among the
variables included in the model at the .05 significance level. Instead of a
vector error correction specification therefore, the series can therefore be
simply modeled in logarithmic first differences.5

when estimating unrestricted, reducedform equations with uniform set


of dependent variables as regressors [Sims (1980), Hamilton (1994)]. This

To test the relationship with one lag, we use a VAR model. Impulse re-

model is appropriate for analyzing the postulated relationships because

sponse functions (IRFs) are generated from the VAR model to analyze

it does not impose a priori restrictions on the structure of the system and

how foreclosures respond to movements in a macroeconomic variable.

can be viewed as a flexible approximation to the reduced form of the

In particular, the IRFs trace the response of one variable to a one stan-

correctly specified, but unknown model of true economic nature. Fur-

dard deviation, one time increase in another variable in the system of

thermore, artificial orthogonal shocks can be introduced in the system

equations and can be thought as a type of dynamic multiplier. The pure

of equations [Doan (1990), Sims (1980), Hamilton (1994)] to capture the

effects of the shocks can then be captured using generalized orthogo-

pure effects and thus assess, in isolation, how sensitive foreclosures

nalization procedures [Runkle (1987)]. However, reporting IRFs without

are to movements in a policy variable.

standard errors or confidence intervals is equivalent to reporting regression coefficient without t-statistics. For statistical inference purposes, we

The VAR model can be expressed as:

Z(t) = C +

use confidence bands around the mean responses based on Doan and
Littermans (1986) Monte Carlo simulation technique. Accordingly, when

A(s)Z(t

s) + e(t) (5)

s=1

the upper and lower bands carry the same sign, the responses are considered to be statistically significant at the 95% confidence level.

where Z(t) is a column vector of variables under consideration, C is the


deterministic component comprised of a constant, A(s) is a matrix of co58

efficients, m is the lag length and e(t) is a vector of random error terms.

5 We nevertheless conducted vector error correction model estimations and found


qualitatively similar results to ones obtained from an unrestricted VAR.

The Capco Institute Journal of Financial Transformation


The US Home Foreclosures and Federal Funds Rate: Is the Channel Effect Muted?

Lastly, Runkle (1987) states that to avoid the problem of over parameterResponse of LOG(HOMIRATE) to LOG(FEDFUND)

ization average situations are models with nine parameters per equation

.04

for 40 data points, 30 for 120 and 70 for 400. The model with one lag in

.02

this study has 64 parameters and 79 observations. However, estimating separate system of equations from one to 11 lags revealed that the
core results are qualitatively the same. To minimize the problem of over
parameterization a decision was made to limit the number of lags to one.

Empirical results

Response of LOG(FEDFUND) to LOG(FORCST)


.4

.2

.00
.0
-.02
-.2

-.04
-.06

10

-.4

Response of LOG(PRICE) to LOG(FEDFUND)

Table 2 reports the coefficient estimates and diagnostic tests for the VAR
variable foreclosure starts have no statistically significant values for the

.00

lag of federal funds rate variable using the conventional significance lev-

-.01

els. Moreover, in the higher lag model the signs of the coefficients at

-.02

first, third, fifth, seventh, ninth, and eleventh lags are negative but not

-.03

significant suggesting that an increase in the federal funds rate does not

10

.10
.05
.00
-.05

10

-.10

Response of LOG(FORCST) to LOG(HOMIRATE)


.15

.04

policy argument, but in favor of the Classical school of thought rather

.10

.02

than Keynesian. Nevertheless, in the VAR model t-stats are relatively

.05

.00

void due to the presence of lagged dependent variables as explanatory

.00

-.02

variables [Sims (1980)]. Instead impulse response functions are utilized

-.05

-.04

to capture the true dynamics of the data generating process, and are

-.10

10

Response of LOG(HOMIRATE) to LOG(FORCST)

have an impact on the foreclosure rate, a finding contrary to effective Fed


6

.15

.02
.01

Response of LOG(FORCST) to LOG(FEDFUND)

.03

model. In the last column, the coefficients belonging to the dependent

10

-.06

10

therefore better suited for statistical inference.7 Thus, the findings from
the impulse response functions are utilized to discuss empirical findings.

Figure 3 Response to Generalized One S.D. Innovations 2 S.E.

Figure 3 plots the impulse response results from the VAR model. The
upper left graph plots the response of the home mortgage interest rates

response is positive and statistically significant up to nine quarters, dis-

(HOMIRATE) to the federal funds rate (FEDFUND). As expected, the

playing persistence. The upper right corner plots the response of the federal funds rate to the foreclosure starts (FORCST). The finding of a lagged
statistically significant negative response is consistent with the behavior
of the Fed and therefore consistent with the view that as foreclosures

Explained variable
FEDFUND

LOG(HOMIRATE)

FEDFUND(-1)

1.045
(0.047)

0.082
(0.024)

0.003
(0.002)

0.012
(0.01)

HOMIRATE(-1)

-0.292
(0.108)

0.794
(0.056)

-0.008
(0.005)

-0.005
(0.022)

LOG(PRICE(-1))

-0.218
(0.336)

-0.199
(0.175)

0.991
(0.015)

0.128
(0.067)

LOG(FORCST(-1))

-0.650
(0.195)

-0.119
(0.101)

-0.008
(0.009)

0.960
(0.039)

2.421
(2.332)

2.024
(1.211)

0.096
(0.103)

-0.715
(0.465)

suggest that foreclosure starts decrease in response to a decrease in the

0.966

0.969

0.996

0.956

with the view that a once and for all rise in the federal funds rate would

600.785

672.298

5326.526

456.608

Akaike AIC

1.310

0.000

-4.923

-1.914

Schwarz SC

1.449

0.138

-4.784

-1.775

R-squared
F-statistic

LOG(PRICE)

( ) denotes standard errors

Table 2 VAR estimation results in logarithmic levels

LOG(FORCST)

increase, the Federal Reserve responds by cutting the federal funds rate.

Explanatory
variable

The middle left graph plots the response of home purchase prices (PRICE)
to the federal funds rate. The lagged negative but statistically insignificant
finding may suggest that even though there is a tendency for prices to fall
in response to a Fed rate increase, the changes are not strong enough to
generate a statistically significant response. The middle right graph is the
core finding which plots the response of the foreclosure starts to the federal funds rate. Unlike the expectation of a positive response which would
federal funds rate, the response is negative but insignificant, consistent

6 The results from estimating separate system of equations from two to 11 lags are available
from the author upon request.
7 The model was estimated by cutting the sample short at third quarter of 2007 to exclude
the recent financial and economic crisis and the results were qualitatively identical.

59

Response of LOG(WEST) to LOG(FEDFUND)

Explained variable

Response of LOG(SOUTH) to LOG(FEDFUND)

.20
.15
.10

Explanatory
variable

D(FEDFUND)

D(HOMIRATE)

D(LOG(PRICE))

D(LOG(FORCST))

.08

D(FEDFUND(-1))

0.623
(0.092)

0.100
(0.056)

0.008
(0.005)

0.002
(0.023)

D(HOMIRATE(-1))

0.081
(0.180)

0.225
(0.109)

-0.024
(0.009)

-0.008
(0.045)

D(LOG(PRICE(-1)))

3.179
(1.919)

3.907
(1.158)

0.174
(0.099)

-0.426
(0.480)

D(LOG(FORCST(-1)))

0.070
(0.425)

0.169
(0.257)

0.044
(0.022)

-0.141
(0.106)

-0.052
(0.045)

-0.077
(0.027)

0.008
(0.002)

0.023
(0.011)
0.024

.04

.05

.00

.00
-.05

.12

-.04
1

10

Response of LOG(NORTHEAST) to LOG(FEDFUND)

10

Response of LOG(NORTHCENTRAL) to LOG(FEDFUND)

.20

.20

.15

.15

.10

.10

.05

.05

.00

.00

R-squared

0.456

0.236

0.128

-.05

-.05

F-statistic

19.032

7.018

3.352

0.564

-.10

-.10

Akaike AIC

0.917

-0.092

-5.009

-1.854

Schwarz SC

1.050

0.041

-4.875

-1.721

10

10

( ) denotes standard errors

Figure 4 Response to Generalized One S.D. Innovations 2 S.E.

Table 3 VAR estimation results in logarithmic first differences

have no statistically significant impact on foreclosures, a finding contrary

for each region to utilize the highest degrees of freedom instead of just

to the effectiveness of traditional Federal Reserve policy argument. This

one VAR system that includes the four regions in one model. Datastream

finding is also supportive of findings from the data section.

reports data from four regions: West, South, Northeast, and North Central. The results are indeed qualitatively similar to findings from estima-

The lower left graph plots the response of the foreclosure starts to home

tions involving national data.9

mortgage interest rates. Consistent with earlier findings, the response is


not statistically significant despite a negative mean response. One would

Figure 4 reports the response of foreclosure starts to the federal funds

naturally conjecture that because the federal funds rate is not able to

rate for the West, South, Northeast, and North Central respectively. Con-

influence the foreclosure starts, home mortgage interest rates which, ac-

sistent with the findings from the national variables, none of the figures

cording to theory, follow the federal funds rate movements do not have

point to a statistically significant response of foreclosure starts to a one

any statistically significant impact on foreclosure starts either.8

standard deviation increase in the federal funds rate. Thus the regional
findings are consistent with the national findings providing further sup-

Lastly, the lower right figure plots the response of the home mortgage

port for the view that the federal funds rate appears to be disconnected

interest rates to foreclosure starts. Consistent with the findings from the

and so consistent with earlier conjectures made.

upper right corner that plots the response of the federal funds rate to the
foreclosure starts, the response of home mortgage interest rates to the

The series in the form of levels contain long memory while in the form of

foreclosure starts is negative and statistically significant but with a lag

differences short memory [Engle and Granger (1991)]. In a short memory

of about five quarters. There is a lagged negative response, because of

system, an old shock to the series has virtually no effect on the current

mostly fixed rate borrowing rather than adjustable rate but the effect is not

value of the series, if the shock happened long enough ago. For a long

channeled back in the form of lower foreclosures rates. This is perhaps due

memory series, an old shock will still have a noticeable impact on the

to the postulated muted effect coming from the federal funds rate. Fur-

current value of the series. The system estimated in levels captures long

ther, when the upper left graph is compared with the lower right graph, the

memory. To capture the short memory dynamics, the series are trans-

home mortgage interest rate appears to respond with a larger lag to move-

formed into logarithmic first differences.

ments in the foreclosure starts than movements to the federal funds rate.
To conduct more in-depth analysis of the data, we use region specific
housing indicators to check consistency with the estimation results from
60

national housing indicators. A separate system of equations is estimated

8 Some practitioners in the field intuitively argue that the home mortgage interest rates are
more responsive to 30 year bond yields.
9 To save space only the impulse responses from the core findings are reported in Figure 4.
The remaining impulse responses are available from the authors.

The Capco Institute Journal of Financial Transformation


The US Home Foreclosures and Federal Funds Rate: Is the Channel Effect Muted?

Response of D(LOG(HOMIRATE)) to D(LOG(FEDFUND))


.04

Response of D(LOG(FEDFUND)) to D(LOG(FORCST))

Response of D(LOG(WEST)) to D(LOG(FEDFUND))

.15

.03

.08

.2
.10

.02

.04

.1

.05

.01

.00

.00

.0

.00
-.01

-.05

10

Response of D(LOG(PRICE)) to D(LOG(FEDFUND))

-.1

10

Response of D(LOG(FORCST)) to D(LOG(FEDFUND))


.12

.03

10

Response of D(LOG(NORTHEAST)) to D(LOG(FEDFUND))

.02

.08

.01

.04

.00

.00

-.04
2

10

Response of D(LOG(FORCST)) to D(LOG(HOMIRATE))

10

.15
.10
.05

.05
.00
-.05

-.05
-.10

10

10

-.10

10

Figure 6 Response to Generalized One S.D. Innovations 2 S.E.

Response of D(LOG(HOMIRATE)) to D(LOG(FORCST))


.04

.12

.08

.03

The finding of no statistically significant response here is as expected be-

.02

cause the system in logarithmic first differences captures short memory.

.01

In a short memory series, an old shock to the series has virtually no effect

.00

on the current value and thus the shock must have happened a short

.04

.00

-.04

Response of D(LOG(NORTHCENTRAL)) to D(LOG(FEDFUND))

.15

.00

-.04

.20

.10

-.01

Response of D(LOG(SOUTH)) to D(LOG(FEDFUND))


.12

.20

.3

10

-.01

10

while ago for the series to respond immediately. Indeed, one would observe in real life that the Federal Reserve response to foreclosures is not
of immediate but of more lagged nature, say after a series of incoming

Figure 5 Response to Generalized One S.D. Innovations 2 S.E.

news about increased foreclosures of a sustained rather than temporary


(once-and-for-all) nature.

Table 3 reports estimation results from a VAR system comprised of the


same variables but in logarithmic first differences to capture any potential

Figure 6 reports regional results showing the response of foreclosure

short run dynamics. Consistent with the findings from the VAR system in

starts to the federal funds rate for the West, South, Northeast, and North

levels, the lag of the differenced federal funds rate does not have a statis-

Central respectively when the system is estimated in the form of logarith-

tically significant impact on foreclosures. Similar to earlier results its own

mic first differences. Consistent with the previous findings, none of the

lag of the foreclosures together with the price series do appear to have a

figures point to a statistically significant response of the foreclosure starts

statistically significant impact at the 10% level.

to a one standard deviation increase in the federal funds rate. The regional findings from a system in first differenced variables again produce

Figure 5 reports impulse responses from the system estimated in loga-

results consistent with the national and regional findings in logarithmic

rithmic first differences. In these estimations as well, the response of the

levels. This provides further evidence in favor of the view that changes in

home mortgage interest rate to federal funds rate is significant but much

the federal funds rate do not appear to have a statistically significant im-

short lived with three quarters lag. As expected, the persistence that ap-

pact on foreclosure starts. This finding is consistent with the view that the

pears in the VAR system in log levels which captures long memory seems

foreclosures are more likely to be an independent process that is likely to

to have disappeared after estimating the system in logarithmic first differ-

continue until all excess inventories are depleted.10,11

ences that capture short memory. The core finding here, the middle right
graph, is consistent with the results from the prior specification. In other
words, when the system is estimated in differenced form, there appears
to be no statistically significant response of foreclosures to the federal
funds rate either. The difference in these results is the response of the
federal funds rate to foreclosures not being statistically significant. So is
the response of the home mortgage interest rates to foreclosures. These
two results were found to be significant in the form of logarithmic levels.

10 The own response of the foreclosures to a standard deviation increase in foreclosures was
significant in all the VAR estimations providing strong support for this claim. Because other
impulse responses provided similar results, these figures are not reported in this study to
save space, but they are available from the authors.
11 To test whether the results are model specific, an ARCH test was conducted on all
variables. None of them appeared to exhibit an ARCH process after the LM test was not
statistically significant at the 5 and 10% significance levels.

61

of precision. In fact, these tools have often been likened to performing

Coefficient

Std. Error

t-Statistic

Prob.

0.118

0.099

1.191

0.237

DU2

0.624

0.796

0.784

0.435

SLD1

-0.002

0.005

-0.415

0.679

SLD2

-0.053

0.019

-2.777

0.007

global factors at play. Securitization and foreclosures appear to have

SLDP1

0.001

0.012

0.059

0.953

a positive relationship during the study sample. However, estimations

SLDP2

0.011

0.044

0.257

0.798

reveal that the foreclosure starts persistently fail to respond to federal

SLDU1

0.000

0.000

1.642

0.104

funds rate even after the inclusion of securitization as an additional vari-

SLDU2

0.000

0.002

-0.100

0.920

able. We test for structural breaks and include a dummy in the determin-

FORCST(-1)

0.534

0.112

4.740

0.000

istic component. However, again the results do not appear to be different

R-squared

0.970

Adjusted R-squared

0.967

Durbin-Watson stat

1.835

DU1

Table 4 OLS estimation with dummy variables

surgery with a sledge hammer rather than a scalpel.


We also include a securitization ratio to capture partially the impact of

from the original estimation.

Concluding remarks
Several academicians and practitioners stated in the past that the channel effect in the US is relatively muted. These statements have been
merely conjecture with no extensive empirical documentation. This study

Table 4 reports the findings from an OLS estimation accounting for a

uses a multivariate model to document the nature of the relationship be-

possible structural break starting from the third quarter of 2006. Intercept

tween the federal funds rate and housing market activity as proxied by

dummy variables are represented by (DU1 and DU2) whereas the slope

the US foreclosure starts. The estimation results from several different

dummies are represented by (SLD1, SLD2) for the federal funds rate,

VAR models utilizing aggregate and regional data to find evidence con-

(SLDP1 and SLDP2) for the home mortgage interest rate, (SLDU1 and

sistent with the view that the channel effect in the US is relatively muted

SLDU2) for home prices. The first order autoregressive term is repre-

whereby any increase in the federal funds rate does not lead to an in-

sented by (FORCST(-1)). The results show that there is no relationship

crease in foreclosure starts.

before the break but a stronger negative and statistically significant relationship after the break despite the significantly low degrees of freedom.

A policy implication of these results is that given the effect of globaliza-

The results are qualitatively the same with the inclusion of the lagged

tion, financial innovation, and deregulation on monetary aggregates, it

dependent variable. They are also the same without the lagged depen-

may be important to transform policy towards non-traditional roles in or-

dent variable in the form of first differences. The statistically significant

der to have a more pronounced and timely outcome on targeted variables

and positive coefficient of the lagged dependent variable yields stronger

and thereby mitigate the muted channel effect. These non-traditional

support for the independent process postulation. These results are also

roles at the micro level are also likely to be more precise than traditional

consistent with the correlations that are run before and after the break.

ones.

This indicates that the negative correlation, if significant, is greater after


the break than before.
Undoubtedly, inclusion of additional key macro variables would be useful. However, due to limited degrees of freedom, our aim in this paper
was to explain much by little using a more parsimonious specification.
However, we estimate three different models which augment the original
model to include changes in GDP as well as Debt to Income Ratio. All
models show no statistically significant response in foreclosure starts to
the federal funds rate even after inclusion of additional macroeconomics
as controls.
While liquidity matters most, because of the effect of globalization, financial innovation and deregulation in the past couple of decades, monetary aggregates such as monetary base, M1 and M2 have become very
blurry over the years. Moreover, traditional tools such as the discount
62

rate and reserve requirement ratio have become less effective in terms

The Capco Institute Journal of Financial Transformation


The US Home Foreclosures and Federal Funds Rate: Is the Channel Effect Muted?

References
























Barberis, N., A. Shleifer, and R. Vishny, 1998, A model of investor sentiment, Journal of
Financial Economics, 49(3): 307-343
Bollerslev, T., 1986, Generalized autoregressive conditional heteroskedasticity, Journal of
Econometrics, 31: 307-327
Bollerslev, T. and J. Wooldridge, 1992, Quasi-maximum likelihood estimation and inference in
dynamic models with time-varying covariances, Econometric Review, 11: 143-172
Charemza, W. W. and D. F. Deadman, 1997, New Directions in Econometric Practice, London:
Edward Elgar Publishing Limited
Daniel, K., D. Hirshleifer, and A. Subrahmanyam, 1998, Investor psychology and security
market under- and over-reactions, Journal of Finance, 53(6): 2143-2184
DeLong, J. B., A. Schleifer, L. H. Summers, and R.J. Waldmann, 1990, The economic
consequences of noise traders, Journal of Political Economy, 98(4): 703-738
Dickey, D. A. and W. A. Fuller, 1979, Distribution of the estimators for autoregressive time
series with a unit root, Journal of American Statistical Association, 74: 427-431
Dickey, D. A. and W. A. Fuller, 1981, Likelihood ratio statistics for autoregressive time series
with unit root, Econometrica, 49: 1057-1072
Diebold, F.X., 1998, The past, present and future of macroeconomic forecasting, The Journal
of Economic Perspectives, 12: 175-192
Doan, T.,1990, Regression Analysis of Time series, Evanston, IL: VAR Econometrics
Doan, T. and R. Litterman,1986, Users manual RATS:Version 2.0., Evanston IL: VAR
Econometrics
Enders, W.,1995, Applied Econometrics Time Series, New York: John Wiley & Sons
Engle, R. F. and C. W. J. Granger, 1991, Long-Run Economic Relationships: Readings in
Cointegration, New York: Oxford University Press
Fraser, P., M. Hoesli, and L. McAlevey, 2008, House price and bubbles in New Zealand,
Journal of Real Estate Finance and Economics, 37: 71-91
Froyen, R. T., 1996, Macroeconomics, Theories & Policies, fifth edition, New Jersey: PrenticeHall
Hakkio, C. S. and C. S. Morris, 1984, Vector autoregressions: a users guide, Research Working
Paper, Federal Reserve Bank of Kansas City, 84-10
Hardouvelis, G., 1987, Reserve announcements and interest rates: Does monetary policy
matter, Journal of Finance, 42: 407-422
Hamilton, J. D.,1994, Time series analysis, Princeton, NJ: Princeton University Press
Hong, H. and J. C. Stein, 1999, A unified theory of underreaction, momentum trading, and
overreaction in asset markets, Journal of Finance, 54(6): 1839-1885
Johansen, S., 1988, Statistical analysis of cointegration vectors, Journal of Economic
Dynamics and Control, 12: 231-54.
Keynes, J. M., 1936, The General Theory of Employment, Interest and Money, London:
Macmillan
Kyle, A. 1985, Continuous auctions and insider trading, Econometrica, 53(6): 1315-1335
Kwon, C. S. and T.S. Shin, 1999, Cointegration and causality between macroeconomic
variables and stock market returns, Global Finance Journal, 10(1): 71 - 81
Litterman, R.B., 1984, Forecasting with Bayesian vector autoregressions five years of
experience, Staff Report, Federal Reserve Bank of Minneapolis, No.95.
Litterman, R. B. and T. M. Supel, 1983, Using vector autoregressions to measure the
uncertainty in Minnesotas revenue forecasts, Quarterly Review, Federal Reserve Bank of
Minneapolis, 7 (2)
Lui, W. N. Strong, and X. Xu, 1999, The profitability of momentum investing, Journal of
Business Finance & Accounting, 26(9): 1043-1091
Lupoletti, W. M. and R. H. Webb, 1986, Defining and improving the accuracy of macroeconomic
forecasts: contributions from a VAR model, The Journal of Business, 59 (2): 263-285
Rubio, M., 2008, Fixed and variable-rate mortgages, business cycles and monetary policy,
Working Paper, Bank of Spain.
Runkle, D. E., 1987, Vector auto regressions and reality, Journal of Business and Economics
Statistics, 5: 437-442
Shiller, R., 1984, Stock prices and social dynamics, Brookings Papers on Economic Activity,
2: 457-498
Simkins, S., 1995, Forecasting with vector autoregressive (VAR) models subject to business
cycle restrictions, Journal of Forecasting, 11: 569-583
Sims, C. A.,1980, Macroeconomics and reality, Econometrica, 48: 1-49
Sims, C. A., 1996, Macroeconomics and methodology, The Journal of Economic
Perspectives, 10: 105-120
Stevenson, A., V. Muscatelli, and M. Gregory, 1988, Macroeconomic Theory and Stabilisation
Policy, London: Philip Allan
Webb, R. H., 1999, Two approaches to macroeconomic forecasting, Economic Quarterly,
Federal Reserve Bank of Richmond, 85(3), Summer

63

Cutting Edge

Sovereign Credit Risk in a Hidden Markov Regime-Switching


Framework. Part 2
Viewing Risk Measures as Information

66

Cutting Edge

Sovereign Credit
Risk in a Hidden
Markov RegimeSwitching
Framework. Part 2
Louise Potgieter Full time MBA student, Cass Business School, City University
Gianluca Fusai Full Professor in Mathematical Finance, DiSEI University of Piemonte
Orientale

Abstract

proving that the methodology can be tractably extended to a

This research applies a discrete-time Markov-modulated

contingent claims approach, and is investigated as a follow-

model to default probability estimation and adapts it to Mer-

up paper to an extensive methodology found in the previ-

tons contingent claims approach, backing the hypothesis

ous edition of the Capco Journal of Financial Transformation

that a regime-switching framework which allows for struc-

(37) [Potgieter and Fusai (2013)]. CDS quotes are used to

tural shifts can substantially improve the underestimation of

calibrate the regime switching model and are then used to

default probabilities associated with the Merton structural

estimate sovereign assets in both developed and emerging

model. The modeling apparatus is applied to sovereign risk

markets.

67

Introduction

net fiscal assets, and other assets minus entities too important to fail.

The recent credit crisis has raised the awareness of investors and regula-

Liabilities are defined as foreign-currency denominated debt plus a local-

tors concerning the appropriate methods for valuation, trading, and risk

currency liability comprised of local-currency debt and base money. Sov-

management of sovereign debt instruments. Many existing models as-

ereign default arises when sovereign assets cannot cover the promised

sume homogenous market conditions and incorporating changes in mar-

payment on foreign currency debt. The default barrier, in our framework

ket regimes or the economic environment due to a credit event appear

the strike of the above mentioned call option, is therefore defined as the

difficult. This motivates the application of an appropriate valuation model

present value of these payments. While the liabilities are known with a

of sovereign debt in a regime-switching framework. We consider an ex-

fair degree of certainty over any given time horizon, the sovereign assets

tension of the Mertons contingent claims approach to the macro level, as

are more uncertain as assets may change for a large number of reasons.

discussed in Gray et al. (2007) and we allow for structural changes and

Three factors therefore drive default: the sovereign asset value, the vola-

regime-switching according to Liew and Sui (2010).

tility of the assets, and the default barrier. The default barrier is defined as
senior foreign-currency denominated debt [Crouhy et al. (2000)].

The model allows for the computation of the term structure of default
probabilities (PDs) and it is calibrated to observed quotes of sovereign

The foreign-currency debt is modeled as default-free value of debt mi-

Credit Default Swaps (CDSs). This allows us to predict a set of observed

nus an implicit put option. Local-currency liabilities are modeled as an

economic balance sheet information including a sovereigns asset value

implicit call option since such liability demonstrates equity-like fea-

(A) and local-currency liability in foreign currency term (LCL). We use the

tures on a sovereign balance sheet. The local-currency multiplied by

proposed model on both developed and emerging markets. The eco-

the exchange rate is considered a market cap of the sovereign in the

nomic intuition behind the regime-switching Markov-modulated model is

international market.

to incorporate the impact of suddenly changing macroeconomic conditions on the sovereign balance sheet information. This is done by assum-

The CCA approach relies on the relationship between balance sheet en-

ing that the volatility of a sovereigns asset value has switching dynamics

tries to extract an implied estimate of sovereign assets by a calibration

and follows a finite-state discrete Markov-chain where the states of the

procedure where the value of the local-currency liability in foreign curren-

chain could represent the states of the economy. The final aim of the re-

cy terms (LCL) is a call option of the sovereigns assets (A) with the strike

search is to establish a link between the credit market and a sovereigns

price as the default barrier (Bf) defined as foreign-currency denominated

balance sheet and to understand if the credit market conveys useful in-

debt. Instead of using balance sheet claims to predict default probabili-

formation to predict economic balance sheet information.

ties according to a CCA model, we perform a reverse engineering procedure by filtering observed market data through the regime-switching

Potgieter and Fusai (2013) provides the key ideas of the model, describ-

model and we try to infer balance sheet information. This allows us to ap-

ing the process of calculating the probability of default on a N-state hid-

preciate the implied value of a sovereigns assets relative to existing debt

den Markov Model, and derives the value for a sovereigns assets and a

that is observable and could signal a looming credit event. A detailed

local-currency liability in foreign currency terms. The calibrated model is

description of the underlying mathematical framework can be found in

used to calculate the price of a standard European call option according

Potgieter and Fusai (2013).

to the contingent claim approach (CCA). The observed local-currency li-

68

ability (LCL) can be compared with the valuation implied in market quota-

As detailed by Potgieter and Fusai (2013), once we have extracted the

tions through the switching regime model. This allows us to understand

default probabilities from quotes on CDS spreads for different maturi-

how much CDSs, conditional to the use of the proposed model, can tell

ties, in our case 1, 2, 3, 4, 5, 7 and 10 years, the calibration consists of

us about the market estimate of the value of sovereign assets. In practice,

inferring the model parameters, iteratively adjusted to best fit the market

there exist significant differences between the two. We perform a detailed

observed default probabilities using non-linear least-squared-error mini-

empirical analysis on a set of countries, representative of advanced and

mization. Then we can use the calibrated model to infer the value of the

emerging economies. The main conclusion it that information on LCL

sovereign assets that are otherwise unobservable. The basic structure of

can be extracted from CDS market quotes across a variety of countries.

the model and formula can be found in Appendix 1.

Extracting sovereign riskiness from a Markovmodulated Merton model

Application to the sovereign case

Gray et al. (2007) extends the Mertons Contingent Claim Approach to the

developed and emerging markets for various maturities. Countries include

macro level to include a sovereign balance sheet analysis. The assets of

South Africa, Brazil, US, Italy, and Germany. The analysis covers a window

a sovereign for the purpose of this approach comprise foreign reserves,

period which differs depending on the availability of market data.

This section reports the results of the empirical analysis performed in both

The Capco Institute Journal of Financial Transformation


Sovereign Credit Risk in a Hidden Markov Regime-Switching Framework. Part 2

The data

calculated by reference to a daily federal funds rate. For Brazil and South

Credit Default Swaps (CDSs) are generally issued for full range of sov-

Africa, interest rates have been obtained using the zero curve as com-

ereign bond issues, more typically for maturities ranging from one to 10

puted by DataStream.

years. Historically, the most actively traded contracts amongst various


sovereign issuers are for maturities T=[1 2 3 5 7 10] years to maturity. The

The data source for all CDS quotes and balance sheet data includes both

CDS historical market quote series vary from country to country, depend-

Bloomberg and DataStream.

ing on the earliest issue in each respective market. Emerging markets


generally have a longer history. Though the frequency of CDS data is

A case of South Africa

obtainable on a daily basis, given that the sovereign balance sheet data is

The data for South Africa includes CDS market quotes averaged on a

available only on a quarterly basis, the calibration considers an averaged

given quarter for a total of 10 observations from March 2010 to June

CDS market quote on the previous quarter to have the same frequency

2012. The estimated parameters calibrated across the entire term struc-

as the balance sheet data.

ture on the last day of each quarter of the estimation period are presented
in Table 1. The first column is the date when the calibration is performed.

For the purpose of this approach, sovereign default arises when sover-

The second and third column present the volatility in state 1 and 2 re-

eign assets cannot sufficiently cover the promised payment on foreign-

spectively, and provide an indication of how the states are defined e.g.,

currency denominated debt. The default barrier is therefore defined as

good or bad. Columns four and five give indication of the probability

the present value of these payments. The default barrier may be defined

of the Markov chain remaining in state 1 and probability of transitioning

as a KMV-like measure (short-term debt plus one-half long-term debt

to state 1 from state 2 respectively. Column six presents the inverse of

plus interest payments up to a certain time) or senior foreign-currency

the leverage parameter i.e., the ratio of the sovereign asset value and the

denominated debt [Crouhy et al. (2000)]. This research adopts the latter

observed distress barrier. An obligor defaults when the value of its assets

definition. Seniority of debt is inferred from examining the behavior of

falls below the value of its liabilities, or equivalent when its inverse lever-

policymakers during stress periods. These efforts make such debt more

age ratio (the ratio of liabilities to assets) falls below one.

senior to junior claims on sovereign local-currency denominated debt.


Local currency liabilities comprising local-currency debt and base money

From the estimated transition probabilities, the probability that the coun-

are modeled as an implicit call option on a sovereigns assets with the

try continues to stay in state 1 is high while the probability of a transition

strike price as the default barrier since such liabilities demonstrate eq-

to state 2 is low. For example, consider the results on June 29, 2012, aj,i

uity-like features on a sovereign balance sheet. The estimation of the

shows transition probability of moving from state i to state j.

sovereign unknown and unobserved asset value can be extracted from


both the calibrated inverse leverage parameter and the observed distress
barrier. We compare the estimation with an observed local-currency li-

s1

s2

a11

a12

S/K

31/03/2010

85.01%

2.21%

0.9869

0.9891

1.89

30/06/2010

62.97%

3.52%

0.9790

0.9837

1.75

data such as the base money, domestic interest rates, foreign interest

30/09/2010

71.01%

4.15%

0.9853

0.9871

1.92

rates, domestic currency denominated debt as the default barrier, and

31/12/2010

70.03%

10.97%

0.9863

0.9874

2.12

forward exchange rates. We calculate the observed local-currency liabil-

31/03/2011

76.19%

3.14%

0.9885

0.9883

1.91

ity using a five year term.

30/06/2011

93.53%

1.46%

0.9704

0.9864

1.98

30/09/2011

86.37%

1.20%

0.9847

0.9842

2.04

Interest rate data include the Eonia (Euro Overnight Index Average) and

30/12/2011

79.21%

2.55%

0.9774

0.9823

2.09

US OIS (US Over-Night Index Swap). The Eonia is an effective overnight

30/03/2012

84.58%

2.48%

0.9839

0.9863

1.91

interest rate computed as a weighted average of all overnight unsecured

29/06/2012

99.90%

0.27%

0.9990

0.9865

1.89

ability in foreign currency terms. Expression (5) given in Appendix 1 is the


relevant one to obtain LCL observed. The inputs require balance sheet

lending transactions in the interbank market. It is one of the two benchmarks (the other one being Euribor) used in the money and capital mar-

Table 1: Estimated parameters in two-state hidden Markov regimeswitching model: South Africa

kets in the Eurozone. The US OIS is an interest rate swap that allows
financial institutions to swap the interest rates they are paying without
having to change the terms of contracts in place with other financial institutions. The fixed rate of OIS is considered less risky than the corresponding interbank rate (LIBOR), as only the net difference in interest
rates is paid at maturity of the swap. In the United States, OIS rates are

To state 1

To state 2

From state 1

0.9990

0.0010

From state 2

0.9865

0.0135

Table 2: Transition probabilities on June 29, 2012: South Africa

69

Calibration

T=1

T=2

T=3

T=4

T=5

T=7

T = 10

This shows that the Markov model captures the persistence in a state;

31/03/2010

1.90%

2.37%

2.19%

2.89%

3.68%

4.53%

5.43%

in this case state 1. How is the state defined then? The two volatility es-

30/06/2010

2.59%

3.07%

2.24%

2.98%

3.84%

4.79%

5.82%

timates provide a clear indication that the level of volatility defines each

30/09/2010

2.10%

2.51%

2.06%

2.73%

3.50%

4.34%

5.25%

state. Given that s1 >> s2, state 1 and state 2 therefore could be classified

31/12/2010

2.06%

2.39%

1.88%

2.53%

3.36%

4.37%

5.54%

31/03/2011

1.40%

1.84%

1.95%

2.87%

3.83%

4.83%

5.85%

30/06/2011

1.48%

2.09%

1.49%

2.68%

3.90%

5.15%

6.43%

30/09/2011

2.70%

3.29%

3.07%

3.97%

4.98%

6.09%

7.27%

30/12/2011

2.73%

3.27%

2.55%

3.52%

4.61%

5.80%

7.06%

30/03/2012

1.82%

2.38%

2.35%

3.37%

4.46%

5.60%

6.77%

29/06/2012

1.75%

2.45%

3.50%

4.64%

5.83%

7.05%

8.30%

Table 3 Calibrated term structure of default probabilities using regimeswitching model: South Africa

as a volatile or bad economy versus a stable or good economy


respectively. The large difference in the volatilities of the two states may
be attributed to the fact that the historical data used in the estimation
period includes a period of market downturn, several corrections and a
strong rally between October 2011 and April 2012. Volatility as measured
by the Johannesburg Stock Exchanges South African Volatility Index
(SAVI) had increased considerably toward the end of the estimation period, albeit below the heights of the 2008 credit crisis.
The inverse leverage ratio generally fluctuates around values not too far
from 2. We recall that a credit event occurs as soon as the inverse leverage ratio approaches a value of 1. The calibrated term structure of
probability of default (PD) is shown in Table 3. The slope of the PD curve
reflects simply a forward expectation of how the default risk is perceived.
Brigo and Tarenghi (2005) highlight that structural models often imply unrealistic short-term credit spreads. There is also empirical evidence that
structural models underestimate the actual probability of default and the
use of regime-switching model intends to improve the structural model
[Leland (2004), Tarashev (2005), Erlwein et al. (2008)]. The calibrated results
indicate that the model resolves the underestimation drawback and that
the PD over short term maturities are all non-zero. Table 3 and Figure 1
illustrate how the default probability moves over time. We observe a wavelike behavior of the PD along the date axis, holding the maturity constant.
The waves appear to peak around June 2010 and July 2011.

Figure 1 Calibrated term structure of default probability: South Africa

We try to understand the information conveyed by the term structure. In late


April 2010, the Greek debt rating was decreased to junk status by Standard
& Poors amidst fears of default by the Greek Government. Thereafter, risks

70

to global financial stability eased off as the economic recovery gained steam.

Distress
Barrier

Implied
Asset
Value

LCL
Observed

LCL
Estimate
(T=5)

Average
LCL
Estimate

31/03/2010

316,610

599,827

64,526

66,600

65,144

Heavy public debt burdens and weak growth prospects in many advanced

30/06/2010

334,956

584,993

62,147

63,232

62,314

economies combined with a series of shocks to the global financial system

30/09/2010

311,111

597,524

75,756

77,231

76,144

were the culprits. Despite higher growth prospects for emerging economies,

31/12/2010

299,561

634,429

81,653

82,607

81,807

markets faced the risk of sharp reversals. At this stage, the crisis moved into

31/03/2011

311,863

596,100

80,162

84,476

82,443

its fifth year, entering into a new phase in which political differences across

30/06/2011

318,272

631,165

83,871

87,120

85,209

economies were impeding progress to address the legacies of the crisis.

30/09/2011

383,228

782,489

76,628

81,232

79,280

An on-going low interest rate environment in developed markets and high

30/12/2011

390,675

814,713

82,327

85,226

83,672

uncertainty drove the asset allocations of institutional investors, with a clear

30/03/2012

375,333

718,745

82,077

88,584

86,039

shift to safety and liquidity. Net capital flows to emerging markets such as

29/06/2012

409,865

773,502

81,336

95,011

90,597

South Africa remained relatively strong during the first half of 2011 although

Table 4 Implied sovereign asset value, distress barrier and localcurrency liability (in millions USD): South Africa

In September 2011, the Global Financial Stability Report cautioned that the
risks to global financial stability increased substantially in prior months.

very volatile. This reflected higher nominal interest rates, the perception that
currencies would appreciate, and relatively strong fundamentals.

The Capco Institute Journal of Financial Transformation


Sovereign Credit Risk in a Hidden Markov Regime-Switching Framework. Part 2

900.000

In order to better quantify this reliability, we next examine the explana-

800.000

tory power for the observed local-currency liability (LCL) over a T= 5 year

700.000

maturity, using the estimated LCL obtained from the model. For this pur-

600.000

pose, we consider the following linear regression where the goodness of


the implied LCL estimate in predicting the observed LCL is tested. The

500.000
400.000
300.000

Distress barrier

relationship to be estimated between the observed and estimated LCL

Implied asset value

is as follows:

200.000
100.000
05/2012

03/2012

11/2011

01/2012

09/2011

07/2011

05/2011

03/2011

01/2011

11/2010

09/2010

07/2010

05/2010

LCLObserved = + b LCLEstimate +
03/2010

(1)

Here represents the intercept of the linear regression, b the slope and
, the error term reflecting other factors that influence the observed local-

Figure 2 Implied sovereign asset value versus distress barrier


(in millions USD) South Africa

currency liability. The estimated relationship turns out to be:


^

LCLObserved = 17 644.63 + 0.73 LCLEstimate


100.000
95.000

with additional outputs given in Table 5.

90.000
85.000

Coefficients

Standard Error

t Stat

P-value

17644.63

8123.17

2.1721

0.061613

0.7322

0.09948

7.3605

7.91E-05

80.000
75.000

LCL estimate (T=5)

70.000

Average LCL estimate

65.000

LCL observed

60.000

R = 0.87
05/2012

03/2012

01/2012

11/2011

09/2011

07/2011

05/2011

03/2011

01/2011

11/2010

09/2010

07/2010

05/2010

50.000

03/2010

55.000

Figure 3 Local-currency liability observed versus estimate


(in millions USD: South Africa

Table 5 Regression analysis for observed local-currency liability (in


million USD): South Africa

A one-tailed t-statistic of 7.4 indicates significance in the estimated


coefficients along with the low p-value << 0.05. Of some relevance is
the high R2 value, well above 0.87, which measures how much the total

Using the estimated parameters from the regime-switching model and

variation of the dependent variable LCLObserved is explained by the re-

the sovereign balance sheet data required for the contingent claim ap-

gression. This suggests that the LCLEstimate implied in the CDS quotes

proach (CCA), an implied asset value is then computed at each estima-

has high power in explaining the LCLObserved. On the other hand, given

tion date. Furthermore, the local-currency liabilities (LCL) estimate is


calculated using as an input the implied asset value. We compare the

the estimated value of b lower than 1 due to the fact that LCLEstimate
> LCLObserved, the prediction appears biased i.e., CDS market quotes

balance sheet estimate with the observed equivalent.

overstate the local sovereign liabilities.

The LCL component is reported for both a T = 5 year maturity and the

Table 6 extends the analysis using the LCL estimates implied by different

LCL component averaged across all maturities T= [1 2 3 5 7 10].

points of the term structure of default probabilities. The short term maturities can provide an even greater predictive value of the observed value of

Figure 2 and Figure 3 provide a visualization of the results as a time series


of each balance sheet component. The implied sovereign asset value lies

LCL. In addition, the prediction extracted from the one year maturity also
appears unbiased ( = 0,b =1). The result implies that the estimated LCL

well above the distress value for all quarters indicating that no default

tend to capture the level of the observed balance sheet data very well.

was imminent in the estimation period, a reasonable observation given


the default probabilities are at most below 3% for a three-year horizon.

Maturity

T=1

T=2

T=3

T=4

T=5

T=7

T = 10

This is also confirmed by the inverse leverage ratio that is well above 1 as

R2

0.99

0.99

0.92

0.90

0.87

0.84

0.80

we have seen from Table 1. The value for the LCL estimate for five-year

0.00

275.05

11059.22

14451.81

17644.63

22212.72

29537.93

maturity tracks the observed component closely, overestimating it within

0.99

0.98

0.86

0.79

0.73

0.68

0.61

a reasonable range, signalling the reliability of the proposed reverse engineering procedure.

Table 6 R2 measure and regression analysis of local-currency liability


estimate across term structure: South Africa

71

A case of Brazil
In the case of Brazil, the estimation period ranges from June 2005 to
June 2012. The estimated parameters calibrated across the entire term
structure are presented in Table 7.
The high values of a11 indicate a clear persistence in state 1, which according to the volatility estimates, is classified as a low volatility state.
The inverse leverage ratio varies between 1.5 and 2.4 throughout the
estimation period, suggesting that in this period, Brazil was in quite a
comfortable situation.
Figure 4 shows the term structure of estimated probability of defaults
(PDs) obtained via the regime-switching model. Default probabilities have
remained relatively stable for shorter maturities while the perceived risk
of longer horizons has steadily decreased over the estimation period:
Figure 4 Calibrated term structure of default probability: Brazil

72

s1

s2

a11

a12

S/K

the probability of default for the 10-year maturity steadily declined from

30/06/2005

24.21%

18.65%

0.9990

0.9990

1.8290

above 15% to below 10%, fluctuating at a moderate level with the eco-

30/09/2005

0.81%

7.95%

0.9990

0.9674

1.5049

nomic environment. Moodys and Fitch have viewed Brazil as showing

30/12/2005

22.65%

19.52%

0.9990

0.9990

2.0049

growing economic resilience with cautious fiscal and monetary policy.

31/03/2006

0.81%

8.76%

0.9986

0.9859

1.5035

30/06/2006

0.80%

11.30%

0.9970

0.9990

1.4981

Table 8 presents the sovereign balance sheet data required for the con-

29/09/2006

0.80%

9.34%

0.9990

0.9899

1.5017

tingent claim approach (CCA) and the local-currency liabilities LCL com-

29/12/2006

0.80%

9.60%

0.9990

0.9990

1.4982

ponent, observed and estimated for a T= 5 year maturity and the coun-

30/03/2007

0.81%

8.38%

0.9986

0.9926

1.5013

terparts averaged across the entire term structure.

29/06/2007

0.81%

9.18%

0.9990

0.9990

1.4985

28/09/2007

0.81%

7.27%

0.9990

0.9883

1.5027

Figure 5 and Figure 6 provide a visual perspective of the quarterly time

31/12/2007

0.80%

8.98%

0.9990

0.9919

1.5003

series of the sovereign balance sheet components. The growing eco-

31/03/2008

0.80%

11.50%

0.9990

0.9981

1.5015

nomic conditions in Brazil are evident by the rise in the indicators. The

30/06/2008

0.80%

7.18%

0.9990

0.9831

1.5022

local-currency liabilities estimate overstates the observed value for peri-

30/09/2008

0.68%

6.59%

0.9854

0.9668

1.4369

ods prior to March 2008, proceeding to understate the observed value

31/12/2008

0.75%

21.05%

0.9756

0.9700

1.2733

thereafter. This implies that CDS quotations, differently from South Africa,

31/03/2009

0.80%

22.04%

0.9829

0.9829

1.2643

at first overstate and then understate local sovereign liabilities. The dis-

30/06/2009

0.80%

11.49%

0.9988

0.9989

1.4965

tress barrier remains well below the implied asset value aside from March

30/09/2009

0.82%

8.03%

0.9990

0.9835

1.5013

2009, where a sharp drawdown is experienced. The financial fragility dis-

31/12/2009

0.82%

7.38%

0.9990

0.9827

1.5025

played here is intimately related to the probability of default which surged

31/03/2010

0.86%

7.79%

0.9990

0.9828

1.5022

around this time period, when the US stock market reached its lowest

30/06/2010

3.38%

8.90%

0.9990

0.9803

1.6681

point since the start of the 2008 recession.

30/09/2010

1.25%

10.03%

0.9990

0.9846

1.7390

31/12/2010

3.88%

11.59%

0.9990

0.9879

1.7863

Next, the quarterly predictive ability for the observed local-currency li-

31/03/2011

9.24%

14.71%

0.9990

0.9912

2.0130

ability (LCL) is examined in Table 9, using the estimated local-currency

30/06/2011

8.12%

19.37%

0.9990

0.9990

2.2920

liability obtained from the model. The regression analysis provides the

30/09/2011

0.81%

19.85%

0.9860

0.9859

1.2963

30/12/2011

0.80%

11.54%

0.9989

0.9985

1.4964

30/03/2012

0.81%

8.16%

0.9990

0.9851

1.5015

29/06/2012

0.85%

8.96%

0.9990

0.9838

1.5018

Table 7 Estimated parameters for Markov regime-switching model: Brazil

following estimated relationship:


^

LCLObserved = -8 9857.77 + 1.25 LCLEstimate


The case of Brazil differs from South Africa in that the implied estimate

The Capco Institute Journal of Financial Transformation


Sovereign Credit Risk in a Hidden Markov Regime-Switching Framework. Part 2

Distress
Barrier

Implied
Asset
Value

LCL
Observed

LCL
Estimate
(T=5)

Average
LCL
Estimate

1.400.000

30/06/2005

191,309

572,777

275,667

303,683

296,816

1.000.000

30/09/2005

183,151

648,459

300,826

339,794

333,364

30/12/2005

169,450

591,055

344,974

397,900

391,717

166,652

625,321

349,438

413,395

407,774

30/06/2006

156,661

564,884

369,277

443,110

437,896

29/09/2006

159,560

598,639

362,863

440,745

434,799

29/12/2006

172,589

681,523

388,391

473,197

467,351

30/03/2007

182,082

675,851

402,677

496,021

490,100

29/06/2007

191,358

882,475

432,881

531,538

525,733

28/09/2007

195,331

749,241

501,279

594,251

588,641

31/12/2007

193,219

757,899

591,676

601,720

596,871

31/03/2008

201,637

740,365

613,378

569,518

564,815

30/06/2008

205,536

754,038

696,894

573,100

569,530

30/09/2008

211,381

732,434

725,278

549,452

544,967

800.000
600.000

Implied asset value

400.000

Distress barrier

200.000
0

06/2005
12/2005
06/2006
12/2006
06/2007
12/2007
06/2008
12/2008
06/2009
12/2009
06/2010
12/2010
06/2011
12/2011
06/2012

31/03/2006

1.200.000

Figure 5 Contingent claim approach (in millions USD): Brazil

1.250.000
1.050.000
850.000

198,340

597,776

518,923

413,486

410,399

LCL estimate (T=5)


Average LCL estimate

250.000

31/03/2009

192,676

420,082

484,679

242,094

238,810

30/06/2009

198,996

712,512

553,688

524,346

521,991

30/09/2009

204,934

786,536

699,871

591,623

589,518

31/12/2009

198,192

769,321

765,835

579,311

577,473

31/03/2010

211,532

794,666

728,538

592,566

590,502

30/06/2010

228,649

888,302

781,736

670,571

668,290

30/09/2010

247,812

933,830

813,936

696,602

694,517

31/12/2010

256,804

924,062

876,930

677,673

675,660

31/03/2011

275,947

1,086,314

922,448

822,211

819,955

LCLEstimate

30/06/2011

291,648

1,143,223

1,018,431

862,636

860,418

2
R = 0.81

30/09/2011

298,219

1,055,958

966,011

773,238

769,524

30/12/2011

298,204

1,193,837

874,191

833,793

830,632

30/03/2012

289,606

1,060,937

900,813

784,285

781,434

29/06/2012

287,529

1,017,470

832,855

745,029

741,555

Table 8 Implied sovereign asset value, distress barrier and localcurrency liability (in millions USD): Brazil

LCL observed

450.000

50.000

06/2005
11/2005
04/2006
09/2006
02/2007
07/2007
12/2007
05/2008
10/2008
03/2009
08/2009
01/2010
06/2010
11/2010
04/2011
09/2011
02/2012

31/12/2008

650.000

Figure 6 Local-currency liability observed versus estimate (in millions


USD): Brazil

Coefficients

Standard Error

t Stat

P-value

-89857.77

68743.39

-1.3072

0.202186

1.2518

0.1161

10.7827

2.75E-05

Table 9 Regression analysis for observed local-currency liability: Brazil


Maturity
R2

T=1

T=2

T=3

T=4

0.89

0.87

0.85

0.91

26229.31

-4255.86

-34375.15

-63074.92

1.07

1.11

1.16

1.21

T=5

T=7

T = 10

0.81

0.87

0.82

-89857.77 -136085.64 -192699.06


1.25

1.34

1.48

Table 10 R2 measure and regression analysis of local-currency liability


estimate across term structure: Brazil

provides an upward bias (b>1). However, R2, is still very high.

The explanatory power of the estimated local-currency liability as measured by the R2 across the entire term structure is presented in Table 10.

The high value of

R2

suggests that the regression model of the estimated

Its high value across maturities implies that the estimated LCL tends to

local-currency liability explains the level in the observed quantity rea-

capture the variation in the observed balance sheet data. For Brazil as

sonably well. A one-tailed t-statistic of 10.8 indicates significance in the

well, predictions extracted from shorter maturities appear to have a very

estimated coefficients together with a P-value < 0.05.

low bias.

73

A case of Italy
Table 11 provides the calibrated parameters for Italy over a time period
that extends from June 2008 to June 2012.
The high transition probability a11 of remaining in state 1 indicates a clear
persistence in state 1 which according to the volatility estimate is clearly
a high volatility state. The time pattern of the ratio S/K signals that Italy
went through a very critical period up to June 2011 followed by the start
of a recovery. This can also be appreciated by observing the calibrated
term structure of the default probabilities in Figure 7.
What is observable is how perceived risk at a shorter time horizon jumped
after April 2011, easing off towards the end of the estimation period. During this time Berlusconis government was in conflict over budget cuts
and the default probability of the one-year horizon reached almost 8%
from lows of 2%. Generally, the perceived riskiness of Italian sovereign

Figure 7 Calibrated term structure of default probability: Italy

debt rose over the entire maturity spectrum. Berlusconis government


resigned in November 2011 and was replaced by one led by the former
European Commissioner Mario Monti. The new government contained
not a single party representative of elected parliamentarians and for this

The most recent risk profile shows a hump: default probabilities first de-

reason was defined as Technocratic. However, it received the full support

crease (for maturities of up to five years) but then increase (conditional on

of The Chamber of Deputies. Financial markets hailed Monti for restor-

no default occurring until 2017). This means that markets are expecting

ing Italys credibility by fixing the budget, starting an ambitious reform

dampened reduced risk of default five years from June 2012 probably

agenda, and for steering the country off the cliff of default. This is evident

due to an ease in political tensions and the reduced risk of deadlocked

in our results: the volatility in the bad state declined steeply from 100%

reform. But over the 10 year term they are anticipating greater risk. Inci-

to a 40% low.

dentally, Moodys rating agency downgraded Italys rating by two notches and subsequently downgraded 10 Italian banks in July 2011. Moodys
cited the usual fears related to the Eurozones debt crisis, along with the
increasing likelihood that greater collective support will be needed for

74

s1

s2

a11

a12

S/K

30/06/2008

99.90%

1.42%

99.90%

99.75%

1.108532

30/09/2008

99.90%

0.40%

99.90%

99.70%

1.028349

31/12/2008

99.90%

0.39%

99.90%

99.75%

1.018536

31/03/2009

99.14%

0.11%

99.90%

99.43%

1.006850

30/06/2009

99.61%

0.25%

99.90%

99.57%

1.020102

30/09/2009

99.90%

0.10%

99.90%

99.70%

1.004666

31/12/2009

99.88%

0.12%

99.90%

99.47%

1.007596

31/03/2010

99.90%

0.10%

99.90%

99.66%

1.005046

30/06/2010

97.77%

1.50%

99.90%

99.90%

1.067598

30/09/2010

95.38%

0.37%

99.90%

99.46%

1.019305

31/12/2010

72.75%

0.20%

99.90%

98.98%

1.011849

31/03/2011

99.90%

0.10%

99.90%

99.71%

1.003614

30/06/2011

92.44%

0.33%

99.90%

99.14%

1.024536

30/09/2011

85.37%

5.80%

99.90%

99.26%

1.275628

30/12/2011

99.90%

11.11%

99.06%

99.77%

1.553669

30/03/2012

41.69%

24.12%

67.70%

79.37%

2.166822

29/06/2012

42.33%

35.93%

54.34%

66.90%

3.086906

Table 11 Estimated parameters for Markov regime-switching model: Italy

Italy. Usually an initial declining term structure denotes a high probability


of default in the short run, due to a very poor economic situation.
Figure 8 and Figure 9 illustrate the quarterly time series of the sovereign balance sheet components. The distress barrier and implied asset
value for periods prior to June 2006 lie precariously close, indicating an
imminent default or risk thereof. Thereafter, the implied asset value experiences a sharp spike, influenced by a sharp increase in the inverse
leverage parameter obtained in the regime-switching model. This in turn
leads to an overestimation in the local-currency liability estimate meaning that CDS quotations overstate local sovereign liabilities. The model
behavior of the observed local-currency liability mimics that observed
when comparing the estimated default probability term structure to the
bootstrapped equivalent.
Table 12 presents the results of the local-currency liability (LCL) estimation and observation for a T= 5 year maturity and the corresponding value
averaged across the entire term structure, together with the observed
distress barrier and the estimated sovereign asset.

The Capco Institute Journal of Financial Transformation


Sovereign Credit Risk in a Hidden Markov Regime-Switching Framework. Part 2

Distress
Barrier

2.500.000

729,202

808,342

222,849

260,498

239,305

770,057

277,605

249,385

221,367

31/12/2008

708,299

721,431

318,999

226,554

202,541

Implied asset value

31/03/2009

753,850

759,014

343,891

256,392

226,451

Distress barrier

30/06/2009

777,003

792,621

307,652

228,342

203,217

30/09/2009

816,334

820,146

326,721

268,907

235,747

31/12/2009

790,676

796,685

287,591

173,806

154,956

31/03/2010

838,388

842,622

361,719

280,812

247,478

30/06/2010

830,750

886,909

366,995

277,538

255,028

30/09/2010

836,599

852,754

356,730

285,615

254,903

31/12/2010

811,807

821,427

312,449

186,951

169,003

31/03/2011

812,795

815,729

355,788

278,443

242,821

30/06/2011

824,494

844,727

328,732

257,137

229,105

30/09/2011

746,986

952,878

388,206

473,777

448,480

30/12/2011

673,391

1,046,227

398,073

478,492

459,768

30/03/2012

677,325

1,467,641

434,547

546,608

515,622

29/06/2012

677,325

2,090,841

387,910

485,534

464,156

06/2012

12/2011

03/2012

09/2011

06/2011

03/2011

12/2010

09/2010

06/2010

03/2010

12/2009

09/2009

06/2009

03/2009

12/2008

09/2008

Average
LCL
Estimate

748,828

500.000

06/2008

LCL
Estimate
(T=5)

30/06/2008

1.500.000

LCL
Observed

30/09/2008

2.000.000

1.000.000

Implied
Asset
Value

Figure 8 Implied sovereign asset value versus distress barrier (in


millions USD): Italy

600.000
500.000
400.000

Table 12 Implied sovereign asset value, distress barrier and localcurrency liability (in millions USD): Italy

300.000

LCL estimate (T=5)


Average LCL estimate

200.000

LCL observed

100.000

Coefficients

Standard Error

t Stat

233434.82

24030.98

9.7139

7.31E-08

0.3467

0.07373

4.7023

0.000283

06/2008
09/2008
12/2008
03/2009
06/2009
09/2009
12/2009
03/2010
06/2010
09/2010
12/2010
03/2011
06/2011
09/2011
12/2011
03/2012
06/2012

LCLEstimate

Figure 9 Local-currency liability observed versus estimate (in millions


USD): Italy

P-value

R2 = 0.60

Table 13 Regression analysis for observed local-currency liability: Italy

Maturity

T=1

T=2

T=3

T=4

T=5

T=7

T = 10

R2

0.47

0.50

0.53

0.56

0.60

0.64

0.69

238544.57 241620.00 241736.09 239143.01 233434.82 213021.46 151497.04


0.18

0.21

0.25

0.30

0.35

0.46

0.69

Table 14 R2 measure and regression analysis of local-currency liability


estimate across term structure: Italy

Next, the quarterly predictive power for the observed local-currency li-

observed local-currency liability, and does however conclude a signifi-

ability (LCL) over a T = 5 year maturity is examined in Table 13, using the

cance in the estimated coefficients.

estimated local-currency liability obtained from the model. The regression analysis produces the following estimates:

Table 14 shows the explanatory power of the estimated local-currency


liability as measured by the R2 across the entire term structure. The re-

LCLObserved = 233434.82 + 0.35 LCLEstimate

sults imply that, differently from the countries considered previously, the
model used to estimate the LCL tend to improve in its ability to capture

The value of R2, the one-tailed t-statistic of 4.7 and the p-value < 0.05,

the variation in the observed balance sheet data as the maturity of the

suggests a moderate linear relationship between the model and the

term structure increases.

75

A case of the United States


For the United States, the estimation period includes quarterly data from
June 2008 to June 2012. The estimated parameters across the entire
term structure are presented in Table 15.
Persistence in state 1 of the model is evident from the results and according to the volatility estimate, it can be classified as a high volatility
state. The difference between s1 and s2 is indeed very large. The inverse
leverage ratio fluctuates around 1.7, reaching levels of 2.4 over the estimation period.
From Figure 10 we observe how the default probability has remained relatively stable for shorter maturities while the perceived risk over a longer
horizon has steadily increased over the estimation period. The probability
of a default (PD) for the 10-year maturity has steadily risen from above
1% to near 4%, fluctuating with the economic environment. We observe

Figure 10 Calibrated term structure of default probability: US

a wave-like behavior of the PD along the date axis, holding maturity constant, particularly pronounced around December 2008 and July 2011.
On 15 June 2011, Dow Jones reported that the one-year CDS spread for

The quarterly time series of the sovereign balance sheet components

the United States was at 43 basis points (higher than the 41 basis points

estimated according to the contingent claim approach for T= 5 year ma-

spread for Brazil), and that the cost of insuring one-year US debt against

turity and the LCL averaged across the term structure are reported in

default had risen on the back of concerns related to the debt ceiling. The

Table 16.

potential of sovereign default or restructuring and concerns that Eurozone fiscal pressures could spread is reflected in the rise in perceived

Figure 11 and Figure 12 serve to illustrate the time series presented in

risk over the longer term.

Table 16. The value for the local-currency liability (LCL) estimate lies above
the observed value indicating that the regime-switching method tends to
overestimate the observed LCL. This means that sovereign CDS quotations on US debt tend to overstate local sovereign liabilities. The implied

76

s1

s2

a11

a12

S/K

sovereign asset value lies well above the distress value for all quarters sug-

30/06/2008

99.90%

0.10%

99.00%

99.76%

1.5847

gesting that no default was imminent in the estimation period. The peak

30/09/2008

81.41%

0.15%

98.57%

99.60%

1.6992

and sudden drop in the implied asset value in 2011 highlight the sensitivity

31/12/2008

72.04%

8.66%

99.90%

99.62%

1.6530

of the approach to the inverse leverage parameters estimated in the model.

31/03/2009

71.35%

8.03%

99.90%

99.63%

1.6572

At this point, the leverage declined, volatility of the model fell sharply, and

30/06/2009

99.90%

0.10%

99.18%

99.65%

1.5893

the transition probability of remaining in a high volatility state declined.

30/09/2009

99.74%

1.91%

99.37%

99.76%

1.5648

31/12/2009

67.95%

8.74%

99.90%

99.80%

1.6565

Next, the quarterly predictive powers for the observed local-currency li-

31/03/2010

69.61%

8.57%

99.90%

99.75%

1.6529

ability (LCL) over a five-year maturity is examined, using the estimated

30/06/2010

68.13%

8.46%

99.90%

99.78%

1.6580

local-currency liability obtained from the model. The regression analysis

30/09/2010

82.98%

9.67%

99.13%

99.64%

1.6361

estimates the following relationship:

31/12/2010

56.23%

0.10%

98.23%

99.21%

1.9686

31/03/2011

92.03%

6.46%

99.06%

99.67%

1.5511

LCLObserved = -654 618.21 + 1.13 LCLEstimate

30/06/2011

81.00%

8.63%

99.90%

99.88%

1.5832

30/09/2011

34.52%

0.10%

98.83%

98.88%

2.3708

The high value of R2 suggests that the regression model of the estimated

30/12/2011

73.38%

8.02%

99.90%

99.84%

1.6024

local-currency liability explains the level in the observed counterpart very

30/03/2012

99.90%

0.10%

99.24%

99.59%

1.5931

well. A one-tailed t-statistic of 7.4 indicates significance in the estimated

29/06/2012

71.40%

9.33%

99.78%

99.66%

1.6570

coefficients. However, it appears biased: implied estimates are higher

Table 15 Estimated parameters for Markov regime-switching model: US

than observed.

The Capco Institute Journal of Financial Transformation


Sovereign Credit Risk in a Hidden Markov Regime-Switching Framework. Part 2

Distress
Barrier

Implied
Asset Value

LCL
Observed

LCL
Estimate
(T=5)

Average
LCL
Estimate

Coefficients

Standard Error

t Stat

-654618.21

358446.60

-1.8263

0.8778

1.1417

0.1543

7.3980

2.23E-06

30/06/2008

21,590,625

34,213,671

741,504

1,828,238

1,640,256

LCLEstimate

30/09/2008

19,204,425

32,633,081

834,246

1,782,926

1,622,492

R2 = 0.79

31/12/2008

19,063,457

31,511,456

1,549,762

1,619,281

1,567,835

31/03/2009

17,755,214

29,423,887

1,580,922

1,878,559

1,818,510

30/06/2009

18,880,466

30,006,668

1,626,160

1,842,701

1,793,222

30/09/2009

20,014,068

31,317,414

1,752,930

1,933,729

1,889,990

31/12/2009

19,730,396

32,682,887

1,952,025

2,121,046

2,075,346

31/03/2010

18,802,148

31,077,619

2,033,102

2,287,998

2,237,842

30/06/2010
30/09/2010

17,112,212
19,624,284

28,371,244
32,108,233

1,977,475
1,925,069

2,211,521

2,168,214

2,252,161

2,181,921
2,223,695

P-value

Table 17 Regression analysis for observed local-currency liability: US

Maturity

T=1

T=2

T=3

T=4

T=5

T=7

T = 10

R2

0.97

0.98

0.97

0.95

0.79

0.64

0.15

473684.52

0.82

143225.44 -188027.20 -489629.72 -654618.21 -249465.89 1405872.59


0.94

1.06

1.14

1.14

0.88

0.18

Table 18 R measure and regression analysis of local-currency liability


estimate across term structure: US
2

31/12/2010

19,349,739

38,091,916

1,970,157

2,283,386

31/03/2011

20,987,132

32,552,364

2,269,313

2,492,781

2,430,674

30/06/2011

21,338,182

33,782,801

2,503,646

2,597,764

2,562,947

30/09/2011

20,031,240

47,489,583

2,589,661

2,786,039

2,751,102

ability as measured by the R2 across the entire term structure. The results

30/12/2011

19,506,948

31,258,363

2,598,679

2,820,027

2,789,482

in fact show an improvement in the predictive power of the model as we

30/03/2012

20,220,053

32,211,718

2,647,675

3,038,320

2,948,079

move toward the short end of the curve.

29/06/2012

19,420,762

32,180,281

2,606,872

3,013,467

2,923,294

Table 16 Implied sovereign asset value, distressed barrier and localcurrency liability (in millions EUR): US

Table 18 shows the explanatory power of the estimated local-currency li-

A case of Germany
In the case of Germany, the estimation period extends from June 2008
to June 2012. The calibrated parameters across the entire term structure
are presented in Table 19.

50.000.000
45.000.000

The striking difference with respect to the other countries is the low vola-

40.000.000

tility levels in the two states, reaching an upper level of around 25%. The

35.000.000
30.000.000
25.000.000

Implied asset value

20.000.000

Distress barrier

15.000.000

high transition probability a11 of remaining in state 1 indicates persistence in a low risk state economy. The inverse leverage parameter ranges
between 1.6 and 2.

10.000.000
0

06/2008
08/2008
10/2008
12/2008
02/2009
04/2009
06/2009
08/2009
10/2009
12/2009
02/2010
04/2010
06/2010
08/2010
10/2010
12/2010
02/2011
04/2011
06/2011
08/2011
10/2011
12/2011
02/2012
04/2012
06/2012

5.000.000

Figure 13 shows the term structure of probability of defaults (PDs) obtained from the estimated PDs. The term structure from the start of the
estimation period until around January 2009 is reasonably flat, indicating

Figure 11 Implied sovereign asset value versus distress barrier (in


millions EUR): US

that perceived risk at the one to 10 year horizon is low and constant. After
this period, the charts show how markets have changed their perception of Germanys long-term prospects over the course of the remaining

3.500.000

period. In two years, the probability of a default over the 10 year horizon

3.000.000

climbs from below 2% to above 6% while the one to three year level re-

2.500.000

mained low. In July 2012, German 10-year debt did, in fact, reach record
lows on the back of fear of a double dip recession in the US. In the same

2.000.000
LCL estimate (T=5)

1.500.000

Average LCL estimate

month, Moodys lowered its outlook on Germany to negative.

LCL observed

1.000.000

The quarterly time series of the sovereign balance sheet components ac-

cording to the contingent claim approach for T= 5 year maturity and the

06/2008
08/2008
10/2008
12/2008
02/2009
04/2009
06/2009
08/2009
10/2009
12/2009
02/2010
04/2010
06/2010
08/2010
10/2010
12/2010
02/2011
04/2011
06/2011
08/2011
10/2011
12/2011
02/2012
04/2012
06/2012

500.000

Figure 12 Local-currency liability observed versus estimate (in millions


EUR): US

LCL averaged across the term structure is reported in Table 20.


Figure 14 shows how the implied sovereign asset value lies well above the
distress barrier for all quarters; analogous to a low probability less than

77

s1

s2

a11

a12

S/K

30/06/2008

2.17%

14.65%

0.9987

0.9985

2.1548

30/09/2008

9.44%

1.33%

0.9982

0.9974

1.4759

31/12/2008

0.60%

13.03%

0.9962

0.9906

1.6380

31/03/2009

1.43%

21.35%

0.9990

0.9949

1.6084

30/06/2009

1.23%

19.16%

0.9990

0.9975

1.5531

30/09/2009

0.10%

20.93%

0.9990

0.9967

1.9859

31/12/2009

0.21%

7.54%

0.9987

0.9951

1.7625

31/03/2010

4.56%

13.97%

0.9988

0.9943

2.0623

30/06/2010

4.65%

7.36%

0.9978

0.9920

1.6989

30/09/2010

3.33%

15.13%

0.9979

0.9932

1.9507

31/12/2010

0.82%

8.22%

0.9974

0.9898

1.4235

31/03/2011

1.97%

11.64%

0.9969

0.9918

2.0313

30/06/2011

0.86%

6.61%

0.9951

0.9938

1.1380

30/09/2011

8.03%

17.86%

0.9986

0.9922

2.3000

30/12/2011

2.78%

20.74%

0.9950

0.9917

1.4242

30/03/2012

0.10%

20.81%

0.9972

0.9921

1.5405

29/06/2012

2.80%

25.17%

0.9933

0.9878

1.8044

Table 19 Estimated parameters for Markov regime-switching model:


Germany

Figure 13 Calibrated term structure of default probability: Germany

8% of Germany defaulting over the estimation period. The implied asset


value also fluctuates considerably from one quarter to the next as ob-

16.000.000

served in Figure 15, analogous to the variability observed in the calibrated

14.000.000

10.000.000
8.000.000

06/2012

04/2012

12/2011

10/2011

02/2012

08/2011

06/2011

04/2011

02/2011

12/2010

10/2010

08/2010

06/2010

04/2010

02/2010

12/2009

10/2009

08/2009

2.000.000

sovereign liabilities.

06/2009

the term structure. This implies that the CDS quotations overstate local

Distress barrier

04/2009

4.000.000

02/2009

the local-currency liability estimate overstates the observed value across

Implied asset value

12/2008

6.000.000

10/2008

observed LCL remains quite stable over time. In practice, we observe that

08/2008

served in the probability of default term structure. On the other side, the

12.000.000

06/2008

inverse leverage parameter and indicative of the wave-like behavior ob-

Next, the quarterly predictive power for the observed local-currency liability (LCL) over a T= 5 year maturity is examined from the regression
results obtained in Table 21, using the estimated local-currency liability

Figure 14 Implied sovereign asset value versus distress barrier (in


millions USD): Germany

obtained from the model. The following relationship is estimated:


^

LCLObserved = -326 508.18 + 0.05 LCLEstimate


The low value of R2 suggests that the regression model of the estimated

900.000
800.000
700.000
600.000

local-currency liability explains very little variation in the observed coun-

500.000

terpart. A one-tailed t-statistic of 0.5 indicates that the estimated coef-

400.000

LCL estimate (T=5)


Average LCL estimate

300.000

ficients are not significant.

LCL observed

200.000

06/2012

04/2012

02/2012

12/2011

10/2011

08/2011

06/2011

04/2011

02/2011

12/2010

10/2010

08/2010

06/2010

04/2010

02/2010

12/2009

10/2009

08/2009

06/2009

04/2009

02/2009

12/2008

10/2008

across the entire term structure. The strik-

08/2008

liability as measured by the

R2

06/2008

100.000

Table 22 shows the explanatory power of the estimated local-currency


ing fact is that the one-year maturity already goes some way to predicting the observed LCL value. In addition, this prediction appears to be
78

unbiased.

Figure 15 Local-currency liability observed versus estimate (in millions


USD): Germany

The Capco Institute Journal of Financial Transformation


Sovereign Credit Risk in a Hidden Markov Regime-Switching Framework. Part 2

Conclusion

Distress
Barrier

Implied
Asset
Value

LCL
Observed

LCL
Estimate
(T=5)

Average
LCL
Estimate

30/06/2008

5,629,811

12,131,255

332,105

766,000

714,543

30/09/2008

5,651,401

8,340,809

296,242

638,475

597,816

31/12/2008

4,869,564

7,976,342

314,663

345,243

337,972

31/03/2009

4,780,107

7,688,443

292,322

390,487

373,137

30/06/2009

4,986,103

7,744,088

304,066

369,280

358,438

30/09/2009

5,172,468

10,272,219

336,526

387,671

378,173

31/12/2009

5,248,690

9,250,694

341,680

386,837

376,706

31/03/2010

5,088,215

10,493,610

337,961

398,169

386,766

30/06/2010

4,915,667

8,351,017

329,289

383,128

372,209

30/09/2010

5,044,933

9,840,971

363,932

438,583

422,975

31/12/2010

5,306,088

7,553,110

408,009

504,486

482,842

31/03/2011

5,218,723

10,600,917

415,524

500,186

480,276

30/06/2011

5,556,759

6,323,703

405,900

486,315

466,626

30/09/2011

5,888,272

13,543,026

367,374

520,247

483,898

The research analyses the economic impact of the estimated default

30/12/2011

5,550,577

7,905,355

352,365

464,870

437,941

probabilities on a sovereigns balance sheet. Using the calibrated regime-

30/03/2012

5,647,322

8,699,482

366,366

510,810

475,596

switching parameters extracted from the Markov-modulated model as

29/06/2012

5,662,888

10,218,330

349,963

524,980

480,244

inputs into the standard option pricing formula in the Merton Contingent

Table 20 Implied sovereign asset value, distressed barrier and localcurrency liability (in millions USD): Germany

Merton-style structural models provide a very appealing feature that links


credit risk with underlying structural variables by providing an endogenous description of credit defaults and an intuitive economic interpretation. However, they suffer the disadvantage of leading to underestimates
particularly over the short term. This research offers the potential to both
resolve the underestimation inherent in most standard structural models
and establish a link between the credit market and a sovereigns balance
sheet in an attempt to understand whether credit markets convey useful
information that predicts economic stability. The methodology and application backs the hypothesis that a regime-switching framework, which
allows for structural shifts, can substantially improve default risk estimators and can be tractably extended to a Contingent Claims Approach in
the case of a sovereign, thereby obtaining a link between the credit market and predictions of a sovereigns balance sheet fundamentals.

Claim Approach, a sovereign asset value and local-currency liability in


foreign currency terms (LCL) is extracted. The value of LCL is a call option of sovereign assets (A) with the strike price as the default barrier (B_f)
defined as foreign-currency denominated debt. The LCL is observable in

Coefficients

Standard
Error

t Stat

P-value

326508.18

43320.09

7.5371

1.78E-06

that under the regime-switching approach, the LCL estimate sufficiently

0.04479

0.08862

0.5054

0.6206

captures the level in the observed LCL for all cases bar Germany. The

LCLEstimate

the market and facilitates a comparison with our estimate. We observe

regression results appear very good, albeit biased given non-zero inter-

R = 0.02

cepts and betas not equal to 1. However, if we consider very short matur-

Table 21 Regression analysis for observed local-currency liability:


Germany

ities, predictions improve considerably, even for Germany. We speculate


on the predictive ability of the model to balance sheet information shortly.

Maturity

T=1

T=2

T=3

T=4

T=5

T=7

T = 10

R2

0.99

0.51

0.16

0.06

0.02

0.00

0.05

0.00

115522.37

253870.81

301201.01

323006.38

350404.59

367717.41

1.00

0.63

0.24

0.11

0.05

-0.01

-0.04

Table 22 R2 measure and regression analysis of local-currency liability


estimate across term structure: Germany

First we identify that the LCL requirement according to the model is higher than the observed value for the case of South Africa, the US, and Germany. Brazil and Italy, during certain periods require a lower LCL estimate
than is observed in the market. Any under- or over-estimation of the LCL
estimate relative to the observed value implies that the corresponding
CDS market quotes, jointly with the model used, under- or over-estimate
local sovereign liabilities respectively. We also observe the relation to the
estimated inverse leverage ratio and the implied asset value. Inverse leverage ratios for the case of Italy are very low, showing signs of imminent
default prior to 2011, therefore resulting in a low estimated value for the
LCL for the relevant sample period. A low predicted inverse leverage parameter (close to 1) signifies imminent default and given the nature of the
valuation of the LCL, significantly draws the required LCL lower than its
observed counterpart.
In the August 2012 monthly bulletin of the European Central Bank (ECB),
the linkages between structural, financial, and fiscal imbalances are cited

79

to have led to the sovereign debt crisis and the fragmentation of the

underestimation rises with a decrease in the true PD and a decrease in

financial markets, highlighting major weaknesses in the institutional set

the sample size [Orth (2011)]. This could affect the regulatory approaches

up of the European Monetary Union [ECB (2012)]. Structural rigidities

to risk management. Basel II states that where limited data are avail-

and a build-up of imbalances resulted in more costly adjustments once

able, a bank must adopt a conservative bias to its analysis, adding to its

the crisis erupted. The financial stability framework struggled to identify

estimates a margin of conservatism related to the likely range of errors

and correct systematic risk prior to the crisis and was equally challenged

this applies equally in the case of the sovereign scenarios that we have

when containing the spread of instability across countries and markets

described.

when the risk materialised.

References
The model provided in this research provides a relative valuation framework for contingent claims on a sovereigns asset, estimating with a
degree of caution, the balance sheet requirements necessary to predict
default. In addition, Gray et al. (2007) suggests that the CCA approach
could have implications on the rapidly growing sovereign wealth funds,
particularly for emerging markets governments that have accumulated
large reserves. Once sovereign risk exposures are calculated in a regime-

Brigo, D. and M. Tarenghi, 2005, Credit Default Swap Calibration and Counterparty Risk
Valuation with a Scenario Credit Default Swaps I: No Counterparty Default Risk, Journal of
Derivatives, 8: 29-40
Clement, D., 2012, Interview with Darrell Duffie, The Region [Interview], 15 June
Crouhy, M., D. Galai, and R. Mark, 2000, A comparative analysis of current credit risk models,
Journal of Banking Finance, 24: 59-117
Duffie, D. and M. Thukral, 2012, Redesigning Credit Derivatives to Better Cover Sovereign
Default Risk, Preliminary Draft, Standford University, available at: http://www.darrellduffie.com/
uploads/working/DuffieThukralMay062012.pdf (accessed 31 August 2012.)

switching framework, new ways of transferring sovereign risk can be


explored and new instruments and risk transfer arrangements can be
developed.
Overall, there are several challenges when applying a contingent claim
approach to a sovereign: the lack of any single dominant model, data re-

quirements that severely limit the scope of application for some models,
and dependence of some models results on distributional assumptions.
Furthermore, that market imperfections exist implies that non-market

measures may add significant power in predicting default and balance


sheet information beyond that possible even with an ideal contingent
claims model. The results conclude that most sovereigns observed LCL

Erlwein, C., R. S. Mamon, and T. K. Sui, 2008, The Pricing of Credit Default Swaps under a
Markov-Modulated Mertons Structural Model, Northern American Actuarial Journal, 12(1):
19-46
European Central Bank (ECB), 2012, Eurosystem Monthly Bulletin, August, available at http://
www.ecb.europa.eu/pub/pdf/mobu/mb201208en.pdf (accessed August 27, 2012)
Gray, D., M. Merton, and Z. Bodie, 2007, Contingent Claims Approach to Measuring and
Managing Sovereign Credit Risk, Journal of Investment Management, 5: 5-28
Leland, H., 2004, Predictions of Default Probabilities in Structural Models of Debt, Journal of
Investment Management, 2: 1-28
Liew, C. C. and T.K. Siu, 2010, A hidden Markov regime-switching model for option valuation,
Insurance: Mathematics and Economics, 47: 374-384
Orth, W., 2011, Default Probability Estimation in Small Samples With an Application to
Sovereign Bonds, Seminar of Economic and Social Statistics, University of Cologne
Potgieter, L. and G. Fusai, 2013, Sovereign Credit Risk in a Hidden Markov Regime-Switching
Framework. Part 1: Methodology, The Capco Institute Journal of Financial Transformation, 37:
99-109
Tarashev, N., 2005, An Empirical Evaluation of Structural Credit Risk Models, Working Paper,
Bank of International Settlements

were estimated well enough by implementing the regime-switching parameters in a contingent claims approach. The regression analysis verifies this. Germany proved to be a case where the explanatory power of
the model is weak and the cause is evident on inspection of the implied

Appendix 1

asset value volatility i.e., volatility implied from the leverage parameter

Applying a sovereign contingent claims approach

and not calibrated explicitly. Gray et al. (2007) attribute asset volatility to

Random fluctuations in the market prices of an entitys assets and liabili-

high levels of foreign exchange price volatility. This highlights the need to

ties together with changes in financial inflows and outflows constitute

address the relationship and causality of asset value volatility and various

balance sheet risk. If the total value of the assets falls below the level of

risk indicators.

promised payment on debt, distress and/or default occurs. The value of


the risky debt is calculated as a default-free value of debt less an implicit

The following issues are left for future research. The research estimates

put option on the underlying assets with the strike equal to the promised

the Markov model when the number of states is two. It would be better,

payments. Equity is modeled as an implicit call option with the same

however, to simultaneously estimate the optimal number of states with

underlying asset and strike. The following balance sheet identity ensues:

other parameters. Standard approaches applied to samples with few defaults pose some serious drawbacks which lead to a high probability of
underestimating the true default probability (PD). These include the obvi-

Asset = Equity + Liability


= Implicit Call Option + Default-Free Debt Implicit Put Option

ous effect of a high uncertainty estimation and skewness of the sampling

80

distribution. Given the small probability of defaults and a small sample

The assets of a sovereign for the purpose of this approach comprise

size observed in sovereign CDS data, any default event leading to a low

foreign reserves, net fiscal assets, and other assets minus entities too

PD estimate under standard approaches is unlikely. The likelihood of

important to fail. Liabilities are defined as foreign-currency denominated

The Capco Institute Journal of Financial Transformation


Sovereign Credit Risk in a Hidden Markov Regime-Switching Framework. Part 2

debt plus a local-currency liability comprised of local-currency debt and

MLC

base money. Sovereign default arises when sovereign assets cannot suf-

- is the base money in foreign-currency terms

ficiently cover the promised payment on foreign currency debt. The de-

rd

- is the domestic interest rate

fault barrier is therefore defined as the present value of these payments.

rf

- is the foreign interest rate

The default barrier may be defined as a KMV-like measure (short-term

Bd

- is the domestic currency denominated debt

debt plus one-half long-term debt plus interest payments up to a cer-

Bd,FC - is the domestic currency denominated debt in foreign currency

- is the base money in local-currency terms

terms

tain time) or senior foreign-currency denominated debt [Crouhy, et al.


(2000)]. When a lender makes a loan to a sovereign, an implicit guarantee

XF

- is the forward exchange rate

of that loan equal to the expected loss of default is created. The ac-

- is the volatility of the sovereign assets

tion of the lender consists of pure default-free lending and bearing a risk
of default by the sovereign. Risky debt can be viewed as a contingent

Equation (2) and (4) are typically used to calculate the unknown and

claim on the (stochastic) sovereign assets. The foreign-currency debt can

unobservable sovereign asset value and asset volatility. The calibrated

therefore be modeled as default-free value of debt minus an implicit put

parameters can be used to obtain sovereign risk measures such as dis-

option. Local-currency liabilities are modeled as an implicit call option

tance-to-default and probability of default and spreads on debt.

since such liability demonstrates equity-like features on a sovereign


balance sheet. Excessive issue of both the money base and local-cur-

Here we follow a different procedure. Instead of pricing CDSs or evalu-

rency liabilities have a similar effect on inflation and price changes as the

ating the value of balance sheet claims according to a CCA model, we

excessive issuing of corporate shares dilute shareholders claims. The

use observed market data, filter it through the regime-switching model

local-currency multiplied by the exchange rate is considered a market

and we try to infer balance sheet information by performing a reverse

cap of the sovereign in the international market.

engineering procedure: we use the calibrated parameters of the regime


switching model to bootstrapped PDs in order to estimate the local-

The main challenge is deriving an estimate for the market value and vola-

currency liabilities in foreign currency terms (LCL) as a call option on a

tility of sovereign assets. Because these are not directly observable, the
CCA approach relies on the relationship between balance sheet entries

sovereigns assets (A) with the strike price as the default barrier (Bf). An
estimation of the sovereign unknown and unobserved asset value (A) can

to extract an implied estimate of sovereign assets by a calibration pro-

be extracted from both the calibrated inverse leverage parameter (S/K)

cedure. The value of the local-currency liability in foreign currency terms

and the observed distress barrier (Bf) such that

(LCL) is a call option of the sovereigns assets (A) with the strike price as
S
K

the default barrier (Bf) defined as foreign-currency denominated debt.


The standard approach requires two equations: the first defines LCL as a

A =

call option on the asset value

The underlying risky asset (S) and the strike (K) in the regime-switching

* Bf 

(6)

framework equates to the sovereign asset value (A) and threshold barrier
LCL = AN(d1) B f erfT N(d2)

(2)

(Bf). Similarly, the inverse leverage parameter (S/K) equates to the ratio of
the sovereign asset value (A) to the default threshold (A/Bf).

(3)

The use of the Merton-type model requires many balance sheet inputs
and parameters which are not always clearly observed and can some-

The second equation defines the volatility of the LCL through

times be inaccurate or difficult to obtain. By reverse engineering the valuation of a sovereigns asset value, the model requires substantially less

LCL*sLCL = AsA N(d1)

(4)

market information and adjusts for any structural breaks in the model in
an attempt to improve the fair value estimates of a sovereigns balance

while sA is the volatility of the sovereigns assets.

sheet.

The formula for the local currency liability in foreign currency which is
observed directly from market data is defined as
LCL = M + Bd,FC =
where

(MLCerdT+Bd)erfT
XF

(5)

81

82

Cutting Edge

Viewing Risk
Measures as
Information
Wayne Tarrant Department of Mathematics, Rose-Hulman Institute of Technology
Dominique Gugan Paris School of Economics, MSE - CES, Universit Paris1, PanthonSorbonne

Abstract
Regulation and risk management in banks depend on underlying risk measures. In general this is the only purpose that
is seen for risk measures. In this paper we suggest that the
reporting of risk measures can be used to determine the loss
distribution function for a nancial entity. We demonstrate
that a lack of sufcient information can lead to ambiguous
risk situations. We give examples, showing the need for the
reporting of multiple risk measures in order to determine a
banks loss distribution. We conclude by suggesting a regulatory requirement of multiple risk measures being reported
by banks, including specic recommendations.

83

Introduction

ten times as much risk in investing $10,000 as there is in investing $1000

In every instance that we can nd, risk measures are used to compute

simply because there is ten times as much money at stake. And most

an amount of capital that an institution should hold in order to remain

would acknowledge that having cash on hand makes them feel safer

nancially solvent, whether this calculation is performed for a banks own

about making an investment. It was exactly these thoughts that were,

internal risk department or for the requirements of an external entity. For

respectively, codied into the denition of risk measure.

instance, regulatory bodies such as the Basel Committee on Banking


Supervision require the reporting of the Value at Risk (VaR) in order to

Let X be a random variable. Then is a risk measure if it satises the fol-

legislate the amount of cash reserves a bank must hold. But should this

lowing properties:

be the only function of a risk measure?

(Monotonicity) if X 0, then (X) 0


(Positive Homogeneity) (X) = (X) 0

In this paper we present the thesis that the reporting of a risk measure

(Translation Invariance) (X + a) = (X) a a R.

is the revelation of a piece of information about the nancial health of


an institution. We show that there are multiple possibilities for loss dis-

The idea behind this denition is that a positive number implies that one

tributions when only a VaR is reported. We then introduce two other

is at risk for losing capital and should have that positive number of a cash

measures, the Expected Shortfall and the Maximum Loss. We show that

balance on hand to offset this potential loss. A negative number would

reporting any one of these measures can lead to multiple possible loss

say that the company has enough capital to take on more risk or to return

distributions.

some of its cash to other operations or to its shareholders.

Following the old adage that more information is always better, we con-

The risk measure that is most used is the Value at Risk (VaR). Essentially

sider the situation of reporting any two risk measures. Again, we show

the -Value at Risk is that number L so that we can expect the losses to

that there is ambiguity about the loss distribution, no matter which two

be worse than L exactly 1 - of the time. For instance, if the 95%-VaR of

measures we choose. We show the same result for three measures and

our position is $100, we would expect to lose more than $100 only 5% of

for ve measures. We then conclude with some suggestions for regula-

the time. A more formal denition can be stated as follows:

tors and with recommendations for further research.

How can we measure risk?

The a- Value at Risk of a position X, VaRa (X) = inf {x : P(X > x) 1 a}

There is little wonder that people are confused by what the term risk

Since the paper of Markowitz (1952), many have also favored the useful-

ought to mean. For instance, the website businessdictionary.com denes

ness of diversication. VaR does not account for this preference, and a

risk in seventeen different general categories. Now, each of the seven-

simple example shows this. Consider the case where a bank has made

teen denitions are relevant to the situations of banks and securities rms.

two $1 million loans and one $2 million loan, each with a 0.04 probability

The unifying theme for each of the denitions is that risk requires both un-

of default and all pairwise independent. Then the 95%-VaR for each loan

certainty and exposure. If a company already knows that a loan will default,

is $0. Thus, if we construct a portfolio consisting solely of the $2 million

there is no uncertainty and thus no risk. And if the bank decides not to loan

loan, we must have a 95%-VaR of $0. If we instead choose diversication

to a business that is considered likely to default, there is also no risk for

and make our portfolio out of the two independent $1 million loans,

that bank as the bank has no exposure to the possibility of loss.

something paradoxical happens. The probability of both loans defaulting


is 0.0016, but the probability of exactly one loan defaulting is 0.0768. This

If our goal is to measure risk, we could attempt to construct a single risk

implies that the 95%-VaR of our diversied portfolio is $1 million. Thus,

measure that will account for all the risk that a bank or securities rm might

there are instances in which VaR does not favor diversication.

encounter. Some have objected to the risk measure being a single number,
but there is some support for this idea. Investing is always a binary deci-

This leads Artzner et al. (1997) to term VaR as incoherent. They dene a

sion: either one invests or one chooses not to invest. Thus the argument is

coherent risk measure as a risk measure that also favors diversication

that, given a single number, one should have enough information to decide

in the following way:

whether to invest or not. There have been some general agreements about

84

the kinds of properties that such a risk measure ought to possess.

if X1, X2 are random variables, then r(X1 + X2) r(X1) + r(X2)

Virtually everyone would agree that if the payout for an investment is al-

The most commonly used coherent risk measure is the Expected Short-

ways positive (after accounting for the risk-free interest rate), then there is

fall. The -Expected Shortfall is the amount one would expect to lose on

truly no risk of loss in the investment. Many people would say that there is

average, given that one is in a case where the loss is greater than the

The Capco Institute Journal of Financial Transformation


Viewing Risk Measures as Information

-VaR. Specically, the -Expected Shortfall of a position X, ES (X), is


ES (X) = E[x]|x < VaR (X).
Finally, most risk managers want to account for the worst case scenario.
Indeed it is only prudent to understand what could be at stake before
agreeing to take a position. This leads to the introduction of one nal risk
measure, called the Maximum Loss. Given a position X, the Maximum
Loss of a position X , ML(X ), is the worst loss one could experience with
non-zero probability, i.e., ML(X ) = inf {(X)|p(X) > 0}.

Ambiguity left by the reporting of risk measures


We note that in the rest of this paper we consider empirical historical risk
measures for which we use a single timeframe for all the risk measures
under consideration. That is to say that all calculations are assumed to be

Figure 1 The region of interest [c,d] in a loss pdf

over the identical historical timeframe. For example, we might consider


a 100-day 95% VaR and a 100-day 95% ES, but in this section we will
not consider a 100-day 95% VaR and a 50-day 95% ES. Of course the
Maximum Loss could be an historical one, though this is often computed
from a model of expected returns or losses.

1%
0

1
%
2

20

Insufciency of a single measure


1%

Here, we discuss the weakness of any single measure of risk for risk
0

management strategy.
In the following we are concerned with tails of probability distribution

10

Figure 2 Loss distributions with 95%-VaRs equal to 0

functions, i.e., if (x) is the pdf of a loss distribution, then we are concerned with the region [c,d] with the property that dc (x)dx = 0.05 and
that (x) = 0 for all x > d. In Figure 1, [c,d] is the region of the graph that we

there are probabilities of a loss in each case. We see that the loss dis-

will consider. This is often called the tail of the loss distribution. In Figure

tribution X1 gives a uniform probability of loss from the VaR of 0 out to

2 and following, we will only see the region [c,d]. Under our notation, the

5. The loss distribution X2 is a very standard tail of a probability distri-

1
.05

dc x (x)dx. We

bution, with the probability of losses decreasing as the size of losses

also want to mention that the probability density function of the loss is

increase. The position X3 is something that might look strange at rst

just the negative of the probability density function for the returns.

glance. However, such loss distributions may occur when risks are highly

95%-VaR(X) = c, the ML = d, and the 95%-ES(X) =

concentrated.
In this and the following sections we will also adopt the notation that the
first graph in each figure represents a loss distribution X1, the second
represents a loss distribution X2, the third a loss distribution X3, and so

For instance, we have seen increasing probability of larger and larger

on. There is no significance to this. It is purely for ease of exposition.

foreclosed upon and sits empty, it is a blight to a neighborhood, causing

losses in the real estate market in some American cities. As one house is
other houses to fall in value and become more likely to enter foreclosure.

Although VaR is used by the Basel Committee for determining the capi-

Since community banks often focus their loans in one particular locality,

talization that a bank needs, it turns out that VaR is insufcient for distin-

there is a potential for such a distribution to occur.

guishing between some kinds of different risk characteristics. Assume that


a bank reports that it has a 95%-VaR of 0 for its present balance sheet.

It is notable that our other two measures can help us distinguish among

5
,
2
20
ES95%(X2) =
= ES95%(X3), so we are unable to tell the difference be3

the three loss distributions. We can calculate that ES95%(X1) =


The bank is able to have any of the tails in Figure 2 as the tail of its loss
distribution.

tween the second and third loss distributions. However, we have that

In each case, the 95%-Value at Risk is 0. This is a great misnomer, as

ML(X1) = 5, ML(X2) = 20, and ML(X3) = 10, which distinguishes each of


these three loss distributions from the other two.

85

5
%
12

1
%
4

16

2
%
3

2
%
3

20

15

15

1
%
3

1
%
3

30

30

Figure 5 Loss distributions with equal 95%-VaRs and equal 95%-ESs

Figure 3 Loss distributions with 95%-ESs of 10

distribution is such that VaR(X1) = 5. For the second and third loss dis-

1%
-5

tributions we know that VaR(X2) = 0 = VaR(X3). And this time the Expected Shortfall sorts out the different distributions because ES(X1) =

2%

5
10
5
, ES(X2) = , and ES(X3) = .
3
3
2

We hope that the geometry of our gures has pointed out some facts
about the different risk measures. In order for us to obtain two distribu-

1%
0

tions with the same 95%-VaR, we only have to begin our loss distribution
5

Figure 4 Loss distributions with MLs of 5

tails at the same value, making sure that we have 5% of the probability
after that common point. For two distributions to have the same 95%-Expected Shortfall, we must have two geometric gures that have the same
weighted average. And for two gures to have the same Maximum Loss,

Whether one argues that certain loss distributions are likely or unlikely,

we would require a common point as the maximum value for which both

there is a stubborn fact that remains. The VaR cannot distinguish among

loss distributions have a non-zero probability, i.e., they need the same

uniform, rising, and falling probabilities. Because most bank regulations

ending value.

only require the reporting of VaR, this should be a very troubling fact.

Insufciency of any two risk measures


Since VaR turned out to be insufcient for our risk measuring purposes,

Since individual measures failed to give a thorough picture of the risk

perhaps one of the two other measures is a better choice. We will start

characteristics of a given loss model, maybe the situation will be better if

with the Expected Shortfall at the 95% level.

we try using two different risk measures. We will attempt to use measures
in pairs to see if they give us a full description of the risk characteristics

In Figure 3 we see that the reporting of the 95%-Expected Shortfall

of a loss model.

leaves the same issue as before. There are uniform, rising, and falling
probabilities, all of which have the same 95%-Expected Shortfall of 10.

We will start by using both the 95%-VaR and the 95%-ES. Again we are

In this case, the Maximum Loss once again identies the three situations

left with a problem. In Figure 5 we see that the loss distribution X1 is uni-

because ML(X1) = 16, ML(X2) = 15, and ML(X3) = 30. Here we also know

form, while X2 has rising probabilities as losses grow, and X3 is the most

that VaR(X1) = 4, while VaR(X2) = 0 = VaR(X3), where each of the VaR cal-

standard looking probability tail. It is discouraging that the two most fre-

culations are at the 95% level.

quently used measures do not distinguish among these three situations.


Each of the distributions has a 95%-VaR of 0 and each of them has a

86

Because the ES has not been able to distinguish among very different

95%-ES of 10. With such different loss characteristics sharing the same

risk situations, we turn to the other measure we have introduced the

VaR and ES, it is astounding that these are the only two measures men-

Maximum Loss. We will again look at the last 5% of the probability dis-

tioned in the Basel III document. We are encouraged, though, because

tribution so that we might be able to consider the usefulness of our other

here the Maximum Loss will distinguish among the loss distributions as

two measures.

ML(X1) = 20, ML(X2) = 15, and ML(X3) = 30.

As we see in Figure 4, the risk models have the same Maximum Loss

Now we try to use the pair of measures that are the 95%-VaR and the

as all the distributions have no probability occurring after the value of

Maximum Loss. As shown in Figure 6, we can still have some ambiguity,

5. Again, there are very different situations of risk here. The rst loss

as in previous examples. In this case we have all three loss distributions

The Capco Institute Journal of Financial Transformation


Viewing Risk Measures as Information

1
%
2

2%

0
0

1%
0

1%
0

10

10

1%
0

2%

10

10

1%
0

5
0

Figure 6 Loss distributions with equal 95%-VaRs and equal MLs

Figure 8 Four loss distributions with equal 95%-VaRs, 95%-ESs and MLs

distribution seems strange, but we argue that it is possible with properly

1
%
4

-5

15
1
%
3

15

-15

chosen correlation. If we go back to the example of community banks


making loans, we would expect a distribution that is very nearly normal.
However, if this bank were to make loans in several communities, it is
possible to see ramping probabilities in one neighborhood due to con-

2
%
3

tagion. Then there may be other neighborhoods where most of the indi15

Figure 7 Loss distributions with equal 95%-ESs and equal MLs

viduals have a large percentage of equity in their house. Such a situation


could lead to the second loss distribution. And similar arguments can apply for the third distribution. Once that nal neighborhood sees the blight
of an entire region due to contagion of foreclosures, we could see that

with 95%-VaRs of 0 and Maximum Losses of 5. Here the 95%-Expected

5
5
Shortfall will distinguish the loss distributions as ES(X1) = , ES(X2) =
,
3
2
10
and ES(X3) = .
3

nal neighborhood also experience increasing numbers of foreclosures.


The fourth loss distribution is nearly impossible for us to justify. However,
as an academic exercise, it does give rise to a potentially innite class

So now we will consider the nal pair, the circumstance of using the

of examples that share the same VaR, ES, and ML. In order for our ex-

95%-ES and the ML as our risk measure pair.

amples to have equivalence of all three measures, we have to meet three


criteria. We need the same starting points of the worst 5% of returns

Once again the three loss distributions shown in Figure 7 have contrast-

in order to make the VaRs equal. We require identical ending points for

ing risk situations, including a uniform probability, one of rising probabili-

the MLs to be equal. Since the uniform distribution has symmetry, it will

ties, and one of declining probabilities. Yet we have each of the three loss

have a 95%-ES of 5. Thus, any distribution with an equivalent 95%-ES

distributions with 95%-ESs of 5 and with MLs of 15. The 95%-VaR does

(and the same starting and ending points) must also be symmetric in

differentiate among the distributions, as VaR(X1) = 5, VaR(X2) = 15, and

the greatest 5% of its losses. So we could form a gure of n triangles of

VaR(X3) = 0.

height 1% and base

Insufciency of all three measures

So our three measures do not distinguish among the potential scenarios

With the failure of two measures to denitively distinguish among differ-

that we might face. Obviously, something more is needed.

10
on the interval [0, 10].
n

ent risk situations, we move to the next logical step. Here we look at the
95%-VaR, the 95%-ES, and the Maximum Loss.

Extending to different probability levels


We have chosen to use the 95% level for our risk measures, but there is

In Figure 8 all four of our examples have a 95%-VaR of 0, a 95%-ES of 5,

nothing special about the 95% level. Our examples would be just as valid

and a ML of 10. The most compelling question here is whether these loss

if we substituted 90% or 99% or any other number in all of our measures

distributions are even possible in actual situations.

and then scaled identically in all of the corresponding diagrams. However, this concept of considering different risk levels might also lead us to

We see uniform distributions with great regularity, so it should be clear

enough risk measures so that we can determine the universal risk prole

that such a distribution is possible in a nancial setting. The second loss

from information about the different risk measures.

87

Conclusion and future directions

1%
0

measures to regulatory bodies, specically a 95%-VaR, 99%-VaR, 95%-

1
%
2

7
2

3
2

ES, and 99%-ES, and a Maximum Loss, where this is possible. We have
demonstrated that the reporting of ve risk measures can theoretically
lead to an innite family of possibilities for the actual loss distribution, but

3
%
2

we have also conceded that such loss distributions are unlikely to occur.

1
%
2

ability in loss distributions when one, two, and three risk measures are
reported. Thus, we suggest that banks be required to submit ve risk

3
%
2

1%

We have shown ambiguity among situations of rising and falling prob-

Thus, we feel that banking regulations would be much safer if such regulations require the reporting of ve different risk measures.

Figure 9 Using five risk measures

This is an interesting theoretical result that must be pursued on actual


loss data. A paper that looks at the ability to differentiate between differSo now we will try using a vector of measures consisting of the 95%-

ent types of common loss distributions has been written with my student,

VaR, the 99%-VaR, the 95%-ES, 99%-ES, and the ML. The different risk

Cole Arendt. I am presently considering the question of how many risk

proles demonstrated in Figure 8 are quite odd-looking, but they all have

measures are needed if we vary the timeframes on the historical data

identical 95%-VaRs, 99%-VaRs, 95%-ESs, 99%-ESs, and MLs.

along with the level and the specic measures. Finally, another paper
that looks at the predictive ability of historical risk measures is also in

The three distributions all have the necessary geometric properties for

progress.

equality of all ve measures.


If we wanted to form a larger family of examples with the same values for
all ve measures, we would need our new distributions to have the same
ending points in order to match up the MLs. We would need them to
have the same starting points for both the greatest 5% of losses and
the greatest 1% of losses in order to have equal 95%-VaRs and equal
99%-VaRs, respectively.
And we would need symmetry in the greatest 5% of losses and in the
greatest 1% of losses in order to have 95%-ESs and 99%-ESs, respectively, equal to the 95%-ES and the 99%-ES of the uniform distribution.

References






Any gure meeting all these criteria would also be indistinguishable from
the rst ve when viewed purely by looking at the values in our vector of ve measures. Clearly several different proles can share all ve

measures.

In spite of our effort in the academic exercise of drawing such gures, we

do not claim that such loss distributions will occur naturally, with the ob-

vious exception of the uniform distribution. Further, the loss distributions

X2 and X3 do not differ by a great deal. Therefore, we are prepared to say

that ve risk measures should distinguish among differing loss distributions in almost all natural circumstances.

88

Artzner, P., F. Delbaen, J. Eber, and D. Heath, 1997, Thinking Coherently, Risk 10: 68-71
Basel Committee on Banking Supervision (BCBS), 1988, International Convergence of Capital
Measurement and Capital Standards, BIS, Basel, Switzerland
Basel Committee on Banking Supervision (BCBS), 1993, Amendment to the Basel Capital
Accord to Incorporate Market Risk, BIS, Basel, Switzerland
Basel Committee on Banking Supervision (BCBS), 2005, Basel II: International Convergence of
Capital Measurement and Capital Standards: A Revised Framework, BIS, Basel, Switzerland
Bassi, F., P. Embrechts, and M. Kafetzaki, 1996, A Survival Kit on Quantile Estimation,
Working Paper, Department of Mathematics, ETHZ, Zurich, Switzerland
Caillault, C. and D. Gugan, 2009, Forecasting VaR and Expected Shortfall Using Dynamical
Systems: a Risk Management Strategy, Frontiers in Finance, 6: 26-50.
Embrechts, P., C. Klppelberg, and T. Mikosch, 2007, Modelling Extremal Events for Insurance
and Finance, Berlin: Springer
Fama, E., The Behavior of Stock Markets, 1965, The Journal of Business, 38: 34-54
Gugan, D., and B. Hassani, 2009, A modied Panjer algorithm for operational risk capital
calculations, Journal of Operational Risks, 4: 1-20
Hogg, R. and S. Klugman, 1984, Loss Distributions, New York: Wiley
Mandelbrot, B., 1963, The Variation of Speculative Prices, The Journal of Business, 36:
394-414
Markowitz, H., 1952, Portfolio Selection, The Journal of Finance, VII: 77-91.
Rockafellar, R. T. and S. Uryasev, 2002, Conditional Value-at-Risk for General Loss
Distributions, Journal of Banking and Finance, 26: 1443-1471
Stulz, R. M., 1996, Rethinking Risk Management, Journal of Applied Corporate Finance, Fall
1996: 8-24
Yamai, Y. and T. Yoshiba, 2002, Comparative Analyses of Expected Shortfall and Value-atRisk: Their Validity under Market Stress, Monetary and Economic Studies Bank of Japan, 20:
181-238

89

Guidelines for Manuscript


Submissions
Capco Journal of Financial Transformation
Guidelines for authors

The first page must provide a short abstract, as well as the full names, titles,

In order to aid our readership, we have established some guidelines to ensure

organizational affiliations and contact details of the authors, including phone

that published papers meet the highest standards of thought leadership and

number and e-mail address.

practicality. The articles should, therefore, meet the following criteria:

Note that the Journal style does not include numbered sections. We
recommend formatting section titles in bold 14 pt, 12 pt and 10 pt to

Does this article make a significant contribution to this field of research?

represent levels of subtitle hierarchy.

Can the ideas presented in the article be applied to current business


models? If not, is there a roadmap on how to get there?
Can your assertions be supported by empirical data?

Bold is used only in subtitles, tables and graphs. Formatting in italics or


underline will not be reproduced.

Is your article purely abstract? If so, does it picture a world that could exist
in the future?
Can your propositions be backed by a source of authority?

Footnotes should be kept to a minimum and be numbered consecutively


throughout the text with superscript Arabic numerals.

Would senior executives find this paper interesting?


Citations in the text should be formatted in the following ways:
Subjects of interest

Author (year) if part of sentence: Smith (2001) mentions that

All articles must be relevant and interesting to senior executives of leading

[Author (year)] if inserted in the sentence: This common approach [e.g.,

financial services organizations. They should assist in strategy formulations.


The topics that are of interest to our readership include:

Smith (2001), Jones (2003)] is the one we used.


[Author (year), page number(s)] if using a direct quote: in the financial
world today [Smith (2001), 25].

Impact of e-finance on financial markets & institutions


Marketing & branding

References at the end of the paper should use the following formats.

Organizational behavior & structure


Competitive landscape

Books

Operational & strategic issues

Copeland, T., T. Koller, and J. Murrin, 1994, Valuation: Measuring and

Capital acquisition & allocation

Managing the Value of Companies, New York: John Wiley & Sons

Structural readjustment
Innovation & new sources of liquidity

e-Books

Leadership

Copeland, T., T. Koller, and J. Murrin, 1994, Valuation: Measuring and

Financial regulations

Managing the Value of Companies, New York: John Wiley & Sons. Kindle

Financial technology

edition
Contributions to collective works

Manuscript guidelines
Manuscripts should be submitted by e-mail directly to Peter Springett:

Ritter, J. R., 1997, Initial Public Offerings, in Logue, D. and J. Seward (eds),
Warren Gorham & Lamont Handbook of Modern Finance, Cincinnati: SouthWestern College Publishing

peter.springett@capco.com.
Journals and periodicals
All submissions must be in US English.

Aggarwal, R. and S. Dahiya, 2006, Demutualization and Cross-Country


Merger of Exchanges, Journal of Financial Transformation, 18: 143-150

Manuscripts should be no longer than 10,000 words, including all footnotes,


references, charts and tables. The manuscript title should be no more than

Government publications

90characters (including spaces).

Bank for International Settlements (BIS), 2011, Quarterly Review, Basel

Manuscripts must be in Microsoft Word, using Times New Roman font, size10.

Newspaper articles

Accompanying the Word document should be a PDF version that accurately

Cumming, F., 1999, Tax-Free Savings Push, Sunday Mail, 4 April, p. 1

reproduces all formulae, graphs and illustrations. When graphs are used,

90

the data should also be provided in a Microsoft Excel spreadsheet format.

Unpublished material

Tables should be in Excel or Word, artwork should be vector-based or in high

Gillan, S. and L. Starks, 1995, Relationship Investing and Shareholder

resolution.

Activism by Institutional Investors. Working Paper, University of Texas

Request for Papers


Deadline November 30, 2013
The world of finance has undergone tremendous change in recent
years. Physical barriers have come down and organizations are
finding it harder to maintain competitive advantage within todays
truly global market place. This paradigm shift has forced managers
to identify new ways to run their operations and finances. The
managers of tomorrow will, therefore, require completely different
skill sets to succeed.

It is in response to this growing need that Capco is pleased to


publish the Journal of Financial Transformation. A journal dedicated
to the advancement of leading thinking in the field of applied finance.

Providing a unique link between scholarly research and business


experience, the Journal aims to be the main source of thought
leadership in this discipline for senior executives, management
consultants, academics, researchers and students. This objective
can only be achieved through relentless pursuit of scholarly integrity
and advancement.

It is for this reason that we have invited some of the worlds


most renowned experts from academia and business to join our
editorial board. It is their responsibility to ensure that we succeed
in establishing a truly independent forum for leading thinking in this
new discipline.

You can also contribute to the advancement of this field by


submitting your thought leadership to the Journal.

We hope that you will join us on our journey of discovery and help
shape the future of finance.

Please contact Peter Springett for more information:


peter.springett@capco.com.

91

Layout, production and coordination: Cypres Daniel Brandt, Kris Van de Vijver
and Pieter Vereertbrugghen
Graphic design: Buro Proper Bob Goor
Photographs: Bob Goor
2013 The Capital Markets Company, N.V.
Prins Boudewijnlaan 43, B-2650 Antwerp
All rights reserved. All product names, company names and registered trademarks in this document remain the property of their respective owners. The
views expressed in The Journal of Financial Transformation are solely those of
the authors. This journal may not be duplicated in any way without the express

92

written consent of the publisher except in the form of brief excerpts or quotations for review purposes. Making copies of this journal or any portion thereof for
any purpose other than your own is a violation of copyright law.

BARUCH COLLEGE
ZICKLIN SCHOOL OF BUSINESS
The largest accredited collegiate school of business in the United States
One of the nations Best Business Schools
Forbes Magazine

#2 in Financial Value at Graduation MBA program


2012 U.S. News & World Report

NATIONALLY RANKED:
#26Graduate Accounting program
#24Undergraduate Accounting program for a second year
2012 Professors Survey Public Accounting Report

www.baruch.cuny.edu/zicklin

Amsterdam
Antwerp
Bangalore
Bratislava
Charlotte
Chicago
Dsseldorf
Frankfurt
Geneva
Hong Kong
Johannesburg
London
New York
Orlando
Paris
San Francisco
Singapore
Toronto
Washington, D.C.
Zurich
capco.com