Credit Portfolio Management
A practioners’ view
Bruno De Cleen
Essex, March 19 2008
20080319_Essex_1_2003.ppt
Introduction
The purpose of this material is to give some input for a discussion about capital and
portfolio management, a domain where theory and practice meet.
The dominant perspective is that of a M, L, XL commercial bank in continental Europe
I would like to demonstrate that things get already complicated before the mathematics
become difficult. Therefore I invite you to put one step back.
I have deliberately left out all the more technical and mathematical aspects.
As this is rather a high level overview, things have been considerably simplified and
therefore may lack accuracy.
The opinions expressed in this document are solely the author’s and do not
necessarily reflect those of my current or previous employers.
Any graphs or tables shown are based on mock data and are for illustrative purposes
only.
20080319_Essex_1_2003.ppt 2
A Commercial Bank (extended)
Executive Board
Audit Finance/Risk
Retail Corporate Asset IT
Management
Private Banking Specialised HR
Finance Real Estate
SME Operations
Corporate Insurance
Publc Sector
Finance
Private Equity
Financial Markets
20080319_Essex_1_2003.ppt 3
The simplified balance sheet
Bank B/S
Property Equity ALM Risk
• Tier 2 (retained earnings) Liquidity Risk
• Tier 2 (Subordinated
Debt) Credit Risk
Loans
Deposits Market Risk
• Without term
• Term
Solvency
Financial Sector Funding
Requrirements
Off Balance Basel 1 – Basel 2
Receivable Liability Target Capital
Rating Agencies
Capital
20080319_Essex_1_2003.ppt 4
All flavours and formats of Credit Risk
Bullet Funded Leverage
Loans
Overdraft facilities
Tranches
‒ Investment
Unfunded ‒ Hedging
Amortising CDS
Guarantees
‒ Wide variety of risks
‒ Rental guarantee
‒ Bidbond,
Revolving ‒ Performance bond
Legal
etc.
Format
Jurisdiction
Counterparty risk for Confirmed vs not
Derivatives confirmed
Immediately
cancellable
20080319_Essex_1_2003.ppt 5
Very different types of exposures
20080319_Essex_1_2003.ppt 6
Geographical Concentration as a rule
20080319_Essex_1_2003.ppt 7
Risk/Solvency Measures
Economic Capital
Expected Loss VAR (1-x bps)
VAR has some conceptual
shortcomings which however in
practice don’t cause too much
problems
Calculation in the tail is difficult
Unexpected Small changes in parameters can
Loss have large impact
Economic Capital
Basel 2
Basel 1 regime for credits was The main perceived shortcomings Transition Regime:
fairly arbitrary (100, 50, 20 % as to solvency for credit risk are: Floor of 90/80% of old Basel 1
weightings) ‒ Imposed correlations regime
Basel 2 (pillar 1), Internal Ratings ‒ Structured products
based Approach, calculates ‒ Assumed single name
solvency requirements in an concentration adjustment
“economic capital’ way. ‒ Unclear “mild crisis’ calibration
‒ Various imposed caps and floors
20080319_Essex_1_2003.ppt 8
Risk Tolerance and relevant risk
measures
Insolvency
Regulator
Different stakeholders have Rating agencies capital)
different focus and different risk Deposit Holders Economic Capital
tolerance
The same amount of loss can
Going Concern Failure
have different impact depending
Solvency insufficiency B1,B2
on the fact whether the loss has
Management Downgrading
been realised inside or outside
Event or series of events hitting
the core business or the
the ” the front page’’
strategic business plan
Inquiry by Stakeholders
communicated to the market
Ex ante loss tolerance tends to
be bigger than ex-post loss Shareholders P&L Volatility
tolerance
Human beings tend to be bad at
handling probabilistic choices.
They prefer real dollars to
wooden dollars. They tend to
have a linear approach.
20080319_Essex_1_2003.ppt 9
Return Calculations
Traditional Raroc Economic Profit
+ Income + Income
- Cost - Cost
- Losses and provisions - Expected Loss
+ Capital Benefit + Capital Benefit
- Taxation - Taxation
- ------------------------ ------------------------
Risk Adjusted Net Profit Risk Adjusted Net Profit Risk Adjusted Net Profit
/ / - Economic capital * Cost of
Capital or Assets Economic capital Captital
=> percentage =>Raroc ---------------------------
percentage =>Economc Profit
amount
Return on Assets
Return on Equity
Return on Capital
Return on Required Equity
‒ B! And B2
In practice many hybrid concepts are used
Different measures are used simultaneously
Even if called the same, calculations can be very different
20080319_Essex_1_2003.ppt 10
The Spectrum of Credit Portfolio
Approaches
Economic/ Regulatory Profit Centre/
Limit Management
Capital Management ‘Credit Treasury’
Name level notional limit Name level notional limit Centralised view of credit risk
management based on rating management supplemented by RAROC based objectives but
grade/ industry economic capital limits recently overlaid by a mark to
Limits set by board level credit RAROC based objectives market view due to IFRS/ fair value
committees Hedge names or sectors that acc.
CDS used to hedge excess breach limits using CDS, Migrate to the ‘optimal’ portfolio
exposure indices or baskets/ nth to default Overlay macro views on credit,
Economic/ regulatory capital CPPI structures
driven securitisations
Exposure management systems Credit portfolio models that Ability to evaluate the impact of
Credit rating tools aggregate and dis-aggregate risk hedging using tranches, proxy-
LGD tools to individual issuers/ sectors hedges, CPPI
Ability to evaluate the impact of Ability to evaluate the MtoM
single name and basket hedges volatility of the hedge book
MtoM/Total Return view
Default view
20080319_Essex_1_2003.ppt 11
The Building Blocks of Credit Risk Modelling
Dependency
• Default correlations almost not directly observable
Independency building blocks
If
not foreseen by construction, a positive definite
matrix is often already quite a performance
What about dependency if things get really sour
Looking for the perfect copula
PD EaD LGD
Probability of default Exposure at Default Loss Given Default
Limited number of observed defaults Randomness Limited to extremely limited number
for some populations Amortising loans versus growing of observations
Lack of consistency portfolios Often bimodal distributions
‒ Default definitions Revolving exposures ‒ Lose a lot
‒ Legacy systems Counterparty risk in derivative ‒ Lose almost nothing
Calibration issues transactions All types of
Consistency in calibration over
different populations
Limited number of years of history
Group effects
20080319_Essex_1_2003.ppt 12
PD Modelling
Rank Ordering Limited to extremely limited
Rank ordering is most of the number of observations
A patchwork of different
time fairly all right
Estimates from models and
populations
Very different techniques used
experts tend rather to converge
Qualitative vs quantitative or
expert based
‒ Producing
‒ PDs
Ratings mapped to PDs
Randomness is quite an
Calibration
obstacle
A small change on the side of
Events perceived as default
the highest PD’s implies a
triggers can vary over different
massive change of the
popoulations
distribution
‒ 90 days pas due in consumer
finance
‒ 1 day past due on an interest
Harmonisation across the
payment in the bond market
portfolio is very hard
Portfolio ‒ Grace period in shipping
Patchwork ...
20080319_Essex_1_2003.ppt 13
Dependency structures: (Proxy)n
Reference
Portfolio Geography Population
Geography
Dependency data mainly
available for US corporate or
for listed corporate.
The Bank’s portfolio is mainly is
European, non listed, with lots
of retail
Data are bucketed along a few
Segment dimensions => does not fit the
Segment
required dimensions
Informationderived not from
defaults but from equity or asset
values
Asset Values+
Equity Values
20080319_Essex_1_2003.ppt 14
When looking at default
statistics, drawing
conclusions is often less
straight forward than one
would have hoped
Randomness
Inconsistency in
methodology used
Correlation impact
15
LGD Modelling
“Technical’ defaults returning to
the healthy portfolio
Often a re-start can be
envisagead;
‒Liquidity/capital injection
Often Bimodal ‒Carving out the unhealthy part of
the business,
Bank loans often with good
collateral as compared to bonds
Independency with PD often
questionable e.g. Asset based
finance
20080319_Essex_1_2003.ppt 16
Credit risk may be fairly complex to
assess
Group Structures
Geography
Company A
Greek Owner
Liberian Flag
Company B Company C Company D Sailing all over the globe
The Main client is a
Russian Company
SPE Joint venture A Collecting Account in
the US
Insurance in the UK
Group Structures are often complex
Unclear where a group starts and
where it ends
Co-debtorship, Guarantees and There is no simple/unique answer to
Letters of intent of all sorts => what to the question what geographical risk
think about PD and LGD ? the ship financing is exposed to.
Ring fenced structures
Difficult to tell what economic sector
20080319_Essex_1_2003.ppt 17
Credit Risk Modelling = looking for the
adequate compromise
Criteria
Business Acceptance
Market consistency
Robustness
Methodological integrity Objectives and Use
Choices Trading
(bps market consistent)
Solvency
Default versus MtoM (order of magnitude)
Horizon Identification of concentration
Point in Time (PIT) versus through Hedging Diversification
the cycle (TTC) Strategic choices
Instruments/Hedges forced in
framework or Portfolio expressed in
Marketable instruments (replicating
portfolio)
20080319_Essex_1_2003.ppt 18
Accounting Complexity: What you see
is what you get?
Main regimes for
assets with credit risk
Fair value through Assets
Balence P&L
Profit & Loss Sheet
Available for Sale/ Assets
Balence P&L Equity
Fair value through Sheet
Equity
- Impairment
Assets
Loans and Balence P&L
receivables Sheet
- Impairment
20080319_Essex_1_2003.ppt 19
Accounting: The Hedging Trap
Economic view
Value of the Value of the Economic
book book value of the
hedged hedged combined
position
Asymmetrical accounting treatment of the book
hedged and the hedging instrument shows increase
Accounting view volatility where it has been economically reduced
Value of the
Assets book
Balence hedged P&L
Sheet
20080319_Essex_1_2003.ppt 20
The Main Levers of Portfolio
Management
Diversification/Overlay
Investments
Derivatives
L/S Beta Overlays
X-Asset class
Front Door Back Door
Limits Securitisation
Pricing Swaps/Pooling
Incentive systems Derivatives
Acquisitions
20080319_Essex_1_2003.ppt 21
Securitisation
Notes
Thedominant factors in the business case are
usually
Super Senior
‒ Funding cost
Reference or
Transferred ‒ Liquidity
Portfolio
‒ Regulatory Capital Relief
The rationale behind regulatory capital arbitrage
AAA is fairly simple: the purpose of the exercise is to
free-up capital at a cost that is lower than the cost
of raising new capital in the market
Risktransfer and economic capital savings are
often difficult to calculate (CDO2)
Mezzanine
No solution for tall trees
Not all types of assets are suitable
Total
Operational complexity / infrasturcture
Portfolio
An extremely wide variety of structures:
‒ Synthetic transactions versus true sales
‒ Synthetic usually partially funded (SPV)
B1 to B2 transition
20080319_Essex_1_2003.ppt 22
Main Hedging/Credit Spread Tools
Instrument Advantages Disadvantage ECap B1 B2 Cost MtoM
Corporate
Single- Isolates credit risk and Basis
between loan
Name Credit Default best tool for ‘tall trees’ and bond underlying
Swaps Liquidity and
transparency
LoanCredit Default More efficient hedging/ Developing market, lack
Swaps ideal for ‘tall trees’ of liquidity
Can be non mark to Non cancellability/ no
market restructuring
CDS indices Highly liquid instruments Cost and roll down
Ability to put on large No capital relief except
macro hedges where names overlap
Swaptions on CDS Potential
for lower cost Singlename
indices and mark to market market is not liquid
compared to CDS
Tranches on CDS Abilityto pinpoint macro Non linear risk
indices hedges to lower cost Impacted by market
Market well balanced technicals
and liquid
Remark: predominantl y a pre-credit crisis view
20080319_Essex_1_2003.ppt
The Complexity of working with credit
derivatives or “Where is Omicron*”
Views
The Hedge stand alone versus the combined position
MtoM view and Accounting view (if different)
The Greeks and the
Basis Risk Operational aspects
like
Delta Spread difference Bond-CDS Timelyevaluation based on
Gamma But more general difference real market prices
etc. between position hedged and
hedging instrument. Operational speed and
Jump to default Most hedges when looked at accuracy
properly turn out to be proxy
Cliff risk? hedges
* Omicron is the undiscovered Greek character feared by all modellers
20080319_Essex_1_2003.ppt 24
What about the universal or ultimate
Credit Portfolio Model?
Credit Portfolio Market Environment
Environment
Market
Intstruments
Portfolio
Patchwork
Specific
Specific
Trading
Overall Specific
Trading
Models
Portfolio Trading
Models
Model Models
Overall view Specific product view
Low frequency of update High frequency of update
Patchwork Focused – Pure
Order of magnitude Bps ambitions
ambitions
How not to get ripped off on the Bps
How not to save Bps while missing the
point on the effectiveness of the hedge
20080319_Essex_1_2003.ppt 25
Important Modelling Choices
One Step Approach
Static
Hedges ?
Force Market
Portfolio
Instruments into
Model Model Logic
Portfolio Patchwork Hedges
Replicating Portfolio
Dynamic
Hedges?
Market
Portfolio Express Portfolio
in Marketable Vaue
Model Instruments
Replicating
Portfolio Patchwork Portfolio Hedges
20080319_Essex_1_2003.ppt
Assessment of Back-end Transactions
is complex
Qualitative
Quantitative or Time
Difficult to Quantify
EVA-Raroc Liquidity - diversification LT effects
‒ The transaction as such ST effects
‒ Alternative use of the space Transition from one regime to
liberated Legal/Documentation another
Taxation
Funding cost Accounting interpretation
Risk Transfer Operational Aspects
Solvency relief Reputational Risk
Reductionof P&L vol
Eg. Reduction of concentrationl
solutions (Back-end – Front End)
Alternative
Business Perspective – Group Perspective
20080319_Essex_1_2003.ppt 27
Concluding remarks
Things get already complicated before the mathematics become difficult. Sometimes
the stochastic differential equations is the funny part of the exercise.
Although I am a great fan of modelling, I would like to invite you to keep your eyes
open for its limitations.
‒ Beware of apparent precision.
‒ Due to lack of data, assumptions (explicit and … implicit) have great impact
‒ Sometimes but fortunately not always, more advanced mathematical techniques
don’t make things any better. Sometimes it is even quite the opposite.
Beware of the cookbook approach. The value is to be found not in the formulas but in
the rigorous thinking.
Don’t underestimate the importance of regulations, accounting and legal.
Communication with all stakeholders is a crucial but a difficult exercise.
A Master in Common Sense and a PhD in Applied Modesty are great qualifications on
top of the necessary degrees in maths, stats, physics, financial engineering or
econometrics.
20080319_Essex_1_2003.ppt 28
Addional Slides
20080319_Essex_1_2003.ppt 29
Capital Management Activities
Credit
The simplified ICAAP View
Capital Capital Adequacy Capital Market Capital Stakeholder
Assessment and Planning Transactions Optimisation Communication
Quantitative Monitoring Regulatory, Issuance Transactions Internal
Assessment of economic and rating - Tier 1 - Risk transfer - Senior management
- Basel I - Tier 2 - Diversification - Other departments
agency capital levels
- Basel II - Risk taking External
Forecasting and
- Rating Agencies Input into strategic - Local Banks
scenario analysis linked - Investors/analysts
- Economic Capital planning process to risk
to strategic planning - Rating agencies,
taking (within Risk
regulators,
appetite)
Optimisation possibilities
Input
Initiative
Steering
CPM Involvement
20080319_Essex_1_2003.ppt 30
Typical CPM roles/activities at the
Group level
Risk Aggregator: Manager of Credit Risk: Centre of Competence
Periodically build and report global Group Develop portfolio optimisation strategies in Collect, roll out and ensure use of consistent
credit exposures order to enhance group risk/return profile CPM best practices within BU CPM
Monitor portfolio developments and report from a neutral perspective (including functions
major changes to management limited investments) Provide selective CPM functionality for
Conduct stress tests on Global level Propose actions if alarming concentrations Business Units without dedicated CPM unit
on Group level occur Represent all CPM activities internally and
Propose, design and manage transactions externally (e.g. in IACPM meetings)
delivered on a group level Represent Business Unit CPM across the
Involvement in strategy and budget group Develop market transaction
process (esp. capital management) competence
Controller Policy Contributor Enabler
Highlight significant breaches of group- Recommend group-wide exposure limits Keep books on the Group level
wide credit risk limits (name, industry & country level) Provides operational and administrative
Propose immediate action when required Input into policy formulation and review support on the group level
in order to shield the portfolio from severe process
value losses Set guidelines for credit risk related
Monitor BU CPM activity stress testing within Bus
Monitor BU origination policies and initiate Contribute to determination of Rabo’s risk
improvements if needed appetite and capital allocation strategy
20080319_Essex_1_2003.ppt 31
Capital management activities both to
secure and to optimise going concern
Limits
Growth
Guideline
STOP
Hedge
Profit
Capital Budgets
Optimise Going Concern Secure Going Concern
20080319_Essex_1_2003.ppt 32