Bank Risk Management

Credit Risk
EKRP 623
Overview

Do they
work?

Why are
How are these
banks BANKS risks managed?
different?

Why does the

How are these
government get
risks measured?
involved?
What risks are
measured?
Banking & trading books
▪ Banking book: includes all ▪ Trading book: includes
securities that are securities which are
▪ not actively traded by the
▪ traded on the market and
institution and
▪ valued by performance of
▪ meant to be held until they mature
market (marked to market)
▪ accounted for in a different way
from those in the…

BANKING BOOK TRADING BOOK

Equities, FX, FI, Property,
Deposits, Loans
Derivatives…
Forward rates
• Forward rates can be derived from spot rates.
• Spot rates are the result of the market participant’s tolerance for risk and
their collective view regarding the future path of interest rates.
• If we assume that these results are purely a function of expectations, we
can use spot rates to estimate the market’s consensus on future interest
rates.
• Theory:
Suppose you have a 2-year horizon and are offered a choice between
locking in a 2-year spot rate of 8.167% or a 1-year spot rate of 4% and
rolling over your investment at the end of the year into another 1-year
security.
Forward rates
• If the difference between 1 year & 2 year rates is purely a function of
expectations regarding future interest rates, you should be indifferent
between these 2 alternatives. i.e. the result should be identical for both
choices, since market participant’s expectations are perfectly factored in
to the 2 year rate.

1 yr spot
1y rate 1y
rate
forward

2y spot rate with annual compounding

Forward rates
• Since we know that under pure expectations, participants expect either
choice to have the same payoff in 2 years:
2
1.08167 = 1.04 × 1 + 1𝑓1

1.081672
− 1 = 1𝑓1 = 12.501%
1.04

• i.e., investors are only willing to accept 4% on the 1-year bond today (when
they could get 8.167% on the 2-year bond today) because they expect to
receive 12.501% on a 1-year bond 1 year from today.
• This expected rate is a forward rate.
Forward rates
• Similarly, we can back other forward rates out of spot rates. Consider, the
1-year rate 2 years from today. Here we have more options: We can invest
in three 1-year bonds, a 2-year bond, then roll over into a 1-year bond, or a
single 3-year investment.

1.123773 = 1.04 × 1 + 1𝑓1 × 1 + 1𝑓2 option 1

1.123773 = 1.081672 × 1 + 1𝑓2 option 2

1.123773
1𝑓2 = 21.296% = 2
−1
1.08167
Computing spot rates from forward rates
• If we are given a set of forward rates, we can compute the various spot
rates.
• Z1= 1f0 (the 1 year forward rate 0 periods from today)
• Given our set of 1 year forward rates derived earlier, calculate Z3, the 3-year
spot rate:

1 + 𝑍3 3 = 1 + 1𝑓0 × (1 + 1𝑓1 ) × (1 + 1𝑓2 )

1
𝑍3 = 1.04 × 1.12501 × 1.21296 ൗ3 − 1
𝑍3 = 12.377%
Computing spot rates from forward rates
6 month (1 period) forward rate beginning 6 years (12 periods)
1𝑓12
from now.

1 year (2 period) forward rate beginning 4 years (8 periods) from

2𝑓8
now.

3 year (6 period) forward rate beginning 2 years (4 periods) from

6𝑓4
now.

4 year (8 period) forward rate beginning 5 years (10 periods) from

8𝑓10
now.
Computing any forward rate
𝑚+𝑡 1/𝑡
(1 + 𝑧𝑚+𝑡 )
𝑡 𝑓𝑚 = −1
(1 + 𝑧𝑚 )𝑚

• Suppose an investor wants to know the 2 year forward rate three years
from now:
4+6 1/4
(1 + 𝑧4+6 )
4𝑓6 = −1
(1 + 𝑧6 )6

• This means that the following rates are needed 𝑍6 (the 3-year spot rate)
and 𝑍10 (the 5-year spot rate)
Computing any forward rate
• Z6 = 4.752%/2 = 0.02375
• Z10 = 5.2772%/2 = 0.026386
1/4
(1.026386)10
4𝑓6 = −1
(1.02376)6

4𝑓6 = 0.030338

• Therefore, the rate is 3.0338% x 2 = 6.0675%

What are these risks?
▪ CREDIT – the risk of loss due to outright default or credit rating
movements
▪ MARKET – the risk of loss due to market price and yield movements
▪ OPERATIONAL – the risk of loss due to fire, acts of god, earthquakes,
theft, fraud, employee pandemics, data entry mis-types, etc
▪ LEGAL – the risk of loss due to mis-selling, discrimination, etc.
▪ LIQUIDITY – the risk of loss due to market illiquidity (assets cannot be
sold at any price – dearth of buyers) or funding illiquidity (lending
markets slam shut)
▪ INTEREST RATE RISK – risk of loss due to deposit/lending rate
mismatches (or rapidly changing, unmanageable mismatches)
Banking & trading books
◼ Banking book: includes all ◼ Trading book: includes
securities that are securities which are
▪ not actively traded by the institution ▪ traded on the market and
and ▪ valued by performance of
▪ meant to be held until they mature market (marked to market)
▪ accounted for in a different way
from those in the…

BANKING BOOK TRADING BOOK

Equities, FX, FI, Property,
Deposits, Loans
Derivatives…
What is credit risk?
The loss due to the possibility that an issuer of a financial
obligation (bond, note, lease, instalment debt) will be unable to
repay interest & principal on a timely basis (i.e. default)
• What factors does default (hence credit risk) depend on?
Exposure
Economic
sector
Maturity
Geographic
distribution
CREDIT
RISK Probability
of loss

Loss given Concentration

default

• And what does “default” mean?

Definition of default
• A default is considered to have occurred with regard to a particular obligor
when either or both of the two following events have taken place:

✓ The obligor is unlikely to pay its credit obligations in full, without recourse to
actions such as realising security (if held)

✓ The obligor is past due more than 90 days on any material credit obligation.
Overdrafts will be considered as being past due once the obligor has
breached an advised limit or been advised of a limit smaller than current
outstandings. (BII para 452)
Loss measurement
• Losses occur when obligors “default” and “fail to pay”
• How are these measured?
• Consider a simple example:
✓ You have lent 1 unit of currency to 100 people, for 1 year

✓ There’s a possibility some will default – assume 5%

✓ Defaulters will still manage to repay something, assume 60%

• How much could you potentially lose at the end of 1y?

That is, 2% of
95 initial exposure
100
5 3

2
Exposure Probability of Loss given
default default
Loss measurement
Exposure Probability of Loss given Total
default default loss

100  5%  (1 – 60%) = 2
• These losses are “expected” – there is no surprise if 2 units are lost at the
end of the year
• Banks and corporates make provisions for these losses
• This suggests the equation for expected losses is:
Expected losses = 𝐸𝐴𝐷 × 𝐿𝐺𝐷 × 𝑃𝐷
• Of course, the PD can vary considerably, so can the LGD and – in some
loan types, so can the EAD
• These values also interact and combine in complex ways to give rise to
unexpected losses
 LOSS MEASUREMENT

Loss given Probability Exposure

default of default at default
EXTERNAL INTERNAL

Rating agencies KMV Logit Factor models

Exposure at default
• An estimation of the extent to which a bank may be exposed to a
counterparty in the event of, and at the time of, that counterparty’s
default.
• A measure of potential exposure (in currency) as calculated by a Basel
Credit Risk Model for the period of 1 year or until maturity whichever is
sooner. Based on Basel Guidelines Exposure at Default (EAD) for loan
commitments measures the amount of the facility that is likely to be
drawn if a default occurs.
• Under Basel II a bank needs to provide an estimate of the exposure
amount for each transaction, commonly referred to as Exposure at Default
(EAD), in banks’ internal systems. All these loss estimates should seek to
fully capture the risks of an underlying exposure
 LOSS MEASUREMENT

Loss given Probability Exposure

default of default at default
EXTERNAL INTERNAL

Rating agencies KMV Logit Factor models

Loss given default
• Foundation approach
✓ Facility-specific: losses influenced by key transaction characteristics (e.g. collateral +
degree of subordination)
✓ Several methods, but most popular are Gross LGD: Total losses/EAD or Total
losses/unsecured portion of credit line

No collateral
✓ Senior claims on corporates, sovereigns & banks: LGD = 45%
✓ Subordinated claims on above: LGD = 75%

Collateral
✓ LGD* = LGD x (E*/ E)
✓ LGD = senior unsecured exp before recognition of collateral (45%)
✓ E = current exposure value (i.e. cash lent or securities lent/posted)
✓ E* effective exposure:
 LOSS MEASUREMENT

Loss given Probability Exposure

default of default at default
EXTERNAL INTERNAL

Rating agencies KMV Logit Factor models

Default and default rates
• How does one measure default rates – or assess the probability of
default?
• This is a complex field, but it is achieved one of two ways
✓ External ratings – provided by rating agencies. Easy to obtain, but some regulators do
not allow them to be used
✓ Internal ratings – using internal credit portfolio models including scorecards, logit
and probit models
• First, consider external ratings and thus the role of rating agencies (plus
the process broadly followed by rating agencies to obtain (ultimately)
probabilities of default)
 LOSS MEASUREMENT

Loss given Probability Exposure

default of default at default
EXTERNAL INTERNAL

Rating agencies KMV Logit Factor models

External ratings: credit rating agencies
• A credit rating is an opinion on the relative ability of an entity or a
transaction to meet financial commitments
• Opinions not facts
• Relative (not absolute) measure of risk
✓ Entities
✓ Transactions
✓ Financial commitments
✓ Local currency rating (Sovereigns) measures ability of government to meet
obligations in terms of its domestic currency and ability to tax and generate revenue
✓ Foreign currency (Sovereigns) ratings assesses likelihood of a depreciating local
currency which would adversely affect government’s ability to pay back foreign
currency obligations.
An experiment
• You have a few €m that you don't need immediately
• Entities wish to borrow this money from you – they need it now

40, wealthy, house Bank – liquid, been around

& car fully paid off, for 50 years, no investments
earning £1m in credit derivatives, in
developed economy

Sovereign (unnamed) Start-up

with dubious company, just
20, Student, no car, no repayment record – discovered oil
house, living with parents, causing headaches in China
big credit card debt for Europe
An experiment
• Who do you lend to?
• How much interest do you charge to the borrowers?
• Why did you make those choices?
• This is exactly the work of CREDIT rating agencies – although they don't
credit rate individuals, they DO credit rate corporates and sovereigns,
financial instruments, financial transactions, etc.
• They provide these credit ratings to the market so others can decide for
themselves where to place their money without having to do exhaustive
(and expensive) research
• Entities "worry" about their credit ratings – it affects their business and the
way they do business CONSIDERABLY
What are credit ratings?
• A credit rating is an opinion on the relative ability of an entity or a
transaction to meet financial commitments
• Opinions not facts
• Relative (not absolute) measure of risk
• Entities rated
• Transactions rated
• Financial commitments rated

Credit-rating agencies wield immense, quasi-governmental power to

determine which companies within the corporate world are
creditworthy and which are not.
Senator Joe Lieberman
What are credit ratings?
• In issuing and maintaining its ratings, Fitch relies on factual information it
receives from issuers and underwriters and from other sources Fitch
believes to be credible
• Fitch conducts a reasonable investigation of the factual information relied
upon by it in accordance with its ratings methodology, and obtains
reasonable verification of that information from independent sources, to
the extent such sources are available for a given security or in a given
jurisdiction
• Users of Fitch’s ratings should understand that neither an enhanced factual
investigation nor any third-party verification can ensure that all of the
information Fitch relies on in connection with a rating will be accurate and
complete
What are credit ratings?
• A Fitch rating is an opinion as to the creditworthiness of a security
• Ratings are inherently forward-looking and embody assumptions and
predictions about future events that by their nature cannot be verified as
facts
• Despite any verification of current facts, ratings can be affected by future
events or conditions that were not anticipated at the time a rating was
issued or affirmed
• The rating does not address the risk of loss due to risks other than credit
risk, unless such risk is specifically mentioned
• Fitch is not engaged in the offer or sale of any security
What are credit ratings?
• A report providing a Fitch rating is neither a prospectus nor a substitute for
the information assembled, verified and presented to investors by the
issuer and its agents in connection with the sale of the securities
• Ratings may be changed, suspended, or withdrawn at anytime for any
reason in the sole discretion of Fitch
• Fitch does not provide investment advice of any sort
• Ratings are not a recommendation to buy, sell, or hold any security
• Ratings do not comment on the adequacy of market price, the suitability of
any security for a particular investor, or the tax-exempt nature or taxability
of payments made in respect to any security
What rating agencies are not

• Regulators
• Auditors
• Advisors (conflicts of interest)
…but these lines are becoming increasingly blurred
Why do they exist?
• Credit ratings are used by investors as an indication of the likelihood
of receiving money back in accordance with the terms on which they
invested
• Implications for economic stability – Argentina, Russia, Greece, Italy,
Spain, the US, Zimbabwe
• Vested interest in governments to involve themselves in these affairs

Raters hold almost biblical authority. There are 2 superpowers in the

world... the US and Moody's Rating Service... and believe me, it's not clear
sometimes who is more powerful.
- Senator Joe Lieberman
Why do they exist?
• The key problem at the heart of credit rating is not the wrong incentives,
though they play a role. It is the credit rating agencies’ pretence (implicit
or explicit) to knowledge that they cannot possibly have
• To be truly useful, a credit rating agency would need the same skills as
Warren Buffett, the greatest investor of all times. That is of course
impossible
• Non-superhuman investors mostly get it wrong; rating agencies are no
different.
• So rather than be granted God-like status – or as evil, Anglo-Saxon
conspirators, they need to be seen as what they really are: generally well-
intentioned scribblers with spreadsheets.
• They should have the right to say what they want – but not to be taken
too seriously.
Rating agencies: who are they?

• More than 2,400 institutions worldwide

• Ratings and analysis track debt covering more than:
• 100 sovereign nations
• 11,000 company issuers
• 25,000 public finance issuers
• 70,000 structured finance obligations
• Employs more than 2,400 people worldwide, ~1,000 analysts.
• Rates 170,000 corporate, government, and structured finance securities
Rating agencies: who are they?
• 6,300 employees
• Located in 21 countries and markets
• Has played a leading role for more than 140 years
• Ratings on US$ 34 trillion of debt issued in 100+ countries
• Issuers and debt obligations of corporations, states and municipalities, financial
institutions, insurance companies and sovereign governments
• Insight into the credit risk of structured finance deals
• Ratings, indices, equity research, risk solutions
❑ S&P U.S. Indices
❑ S&P/Citigroup Global Equity Indices
❑ S&P Emerging Market Indices
❑ S&P Alternative Indices
Rating agencies: who are they?

• Dual-headquartered in NY & London

• 90 countries 1,500 employees
❑ 3,100 financial institutions
❑ 1,600 banks
❑ 1,400 insurance companies.
❑ 1,200 corporates, 89 sovereigns, 45,000 municipal transactions.
• 8,600 structured finance transactions under surveillance, including
4000 RMBS pools, 440 CMBS, 1600 ABS, and 600 CDOs.
• 1200 European & 200 Asian structured finance transactions.
The rating process

Issuer requests a Rating team visits Issuer makes

rating issuer presentation

Rating committee discusses data and Rating team

votes on rating analyse data

Notification of
rating to issuer

Formal rating Rating released to Ongoing rating agency

notification capital markets monitoring
Ratings
Grade Moody's S&P Fitch Support
Prime Aaa AAA AAA
High grade Aa1 AA+ AA+
Aa2 AA AA

INVESTMENT
Aa3 AA- AA-
Upper medium grade A1 A+ A+
A2 A A 1
A3 A- A-
Lower medium grade Baa1 BBB+ BBB+ 2
Baa2 BBB BBB
Baa3 BBB- BBB- 3

SPECULATIVE
Non-investment grade speculative Ba1 BB+ BB+
Ba2 BB BB 4
Ba3 BB- BB-
Highly speculative B1 B+ B+ 5
B2 B B
B3 B- B-
Subtantial risks Caa1 CCC+ CCC+
Caa2 CCC CCC
Caa3 CCC- CCC-
Extremely speculative Ca CC CC

JUNK
C
In default with little prospect for recovery SD RD
In default C D D
DD
DDD
Not rated WR NR
What do these ratings mean?
AAA: Highest credit quality – the lowest expectation of default risk. Assigned only in cases of
exceptionally strong capacity for payment of financial commitments, highly unlikely to be adversely
affected by foreseeable events
AA: Very high credit quality – expectations of very low default risk. Very strong capacity for payment
of financial commitments, not significantly vulnerable to foreseeable events
A: High credit quality – expectations of low default risk. Capacity for payment of financial
commitments is strong, but may, nevertheless, be more vulnerable to adverse business or economic
conditions than is the case for higher ratings
BBB: Good credit quality - expectations of default risk are currently low. The capacity for payment of
financial commitments is considered adequate but adverse business or economic conditions are more
likely to impair this capacity
BB: Speculative – elevated vulnerability to default risk, particularly in the event of adverse changes in
business or economic conditions over time; however, business or financial flexibility exists which
supports the servicing of financial commitments
B: Highly speculative – material default risk is present, but a limited margin of safety remains.
Financial commitments are currently being met; however, capacity for continued payment is
vulnerable to deterioration in the business and economic environment
What do these ratings mean?
CCC: Substantial credit risk – default is a real possibility
CC: Very high levels of credit risk – default of some kind appears probable
C: Exceptionally high levels of credit risk – default is imminent or inevitable, or the issuer is in
standstill. Conditions that are indicative of a 'C' category rating for an issuer include:
a. the issuer has entered into a grace or cure period following non-payment of a material financial
obligation
b. the issuer has entered into a temporary negotiated waiver or standstill agreement following a
payment default on a material financial obligation; or
c. Fitch Ratings otherwise believes a condition of 'RD' or 'D' to be imminent or inevitable, including
through the formal announcement of a distressed debt exchange
RD: Restricted default – indicate an issuer that has experienced an uncured payment default on a
bond, loan or other material financial obligation but which has not entered into bankruptcy filings,
administration, receivership, liquidation or other formal winding-up procedure, and which has not
otherwise ceased operating
D: Default – indicate an issuer that has entered into bankruptcy filings, administration, receivership,
liquidation or other formal winding-up procedure, or which has otherwise ceased business
Support ratings
1 – an extremely high probability of external support. The potential provider of support is very highly
rated in its own right and has a very high propensity to support the bank in question
2 – a high probability of external support. The potential provider of support is highly rated in its own
right and has a high propensity to provide support to the bank in question
3 – a moderate probability of support because of uncertainties about the ability or propensity of the
potential provider of support to do so
4 – a limited probability of support because of significant uncertainties about the ability or propensity
of any possible provider of support to do so
5 – a possibility of external support, but it cannot be relied upon. This may be due to a lack of
propensity to provide support or to very weak financial ability to do so
Default probabilities (historical long term)

60%
Probability of default (implied)

50%

40%

30%

20%

10%

0%
AAA AA+ AA AA- A+ A A- BBB+ BBB BBB- BB+ BB BB- B+ B B- CCC+ CCC

Credit rating
Why do institutions care about their rating?
• Credit ratings are used by investors as an indication riskiness
• Hence, direct relationship between an institution’s rating and the
amount of interest they must pay on their loans
• Cliff effects, reputational risks, ‘bank runs’
Rating process: economic environment
GDP growth
Size Inflation

Growth in
National
ECONOMIC consumer
property price
ENVIRONMENT lending, savings
indices
and investment

Bond Unemployment
yields Exchange trends
rates
Rating process: operating environment

Structural
changes in Effect of
Consolidated economy
supervision adopting IFRS
directives accounting

Existing and
OPERATING Political
potential
ENVIRONMENT stability
competition

Aggressiveness
Implementation
Oversight of of accounting
of Basel II/III
national figures
regulators
Key topics

Capital
Funding and Performance
liquidity and earnings

Audit
RISK Securitisation

Corporate
Market
governance & Management
environment
ownership strategy
Key topics

CREDIT
MARKET OPERATIONAL

LIQUIDITY RISK STRATEGIC

INTEREST LEGAL
RATE BUSINESS
 Category Overall Sub factor Overall
DEVELOPED weight weight weight weight
Market share and sustainability 20% 2.4%
Geographical diversification 20% 2.4%
Franchise value 40% 12% Earnings stability 20% 2.4%
Earnings diversification 20% 2.4%
Vulnerability to "event" risk 20% 2.4%
Corporate governance 16.7% 2.0%

Qualitative factors
Controls 16.7% 2.0%
Financial reporting transparency 16.7% 2.0%
Risk positioning 40% 12% Credit risk concentration 16.7% 2.0%
30% Liquidity management 16.7% 2.0%
Market risk appetite 16.7% 2.0%
Licensing 10% 0.3%
Capital regulation 15% 0.5%
Regulatory environment 10% 3% Other prudential regulation 15% 0.5%
Supervision 30% 0.9%
Independence enforcement 30% 0.9%
Economic stability 33.3% 1.0%
Operating environment 10% 3% Integrity and corruption 33.3% 1.0%
Legal system 33.3% 1.0%
PPP% average RWA 50% 5.5%
Financial fundamentals

Profitability 15.75% 11%

Net income % average RWA 50% 5.5%
Liquidity 15.75% 11% (Market funds-Liquid assets) % total assets 100% 11.0%
Tier 1 ratio 50% 5.5%
70%

Capital adequacy 15.75% 11%

Shareholders equity % total assets 50% 5.5%
Efficiency 7% 5% Cost/income ratio 100% 4.9%
Problem loans % gross loans 50% 5.5%
Asset quality 15.75% 11%
Problem loans% (equity + LLR) 50% 5.5%
Lowest score 30% 21% Assigned to lowest scoring financial ratio 100% 21.0%
 Category Overall Sub factor Overall
DEVELOPING weight weight weight weight
Market share and sustainability 20% 1.4%
Geographical diversification 20% 1.4%
Franchise value 10% 7% Earnings stability 20% 1.4%
Earnings diversification 20% 1.4%
Vulnerability to "event" risk 20% 1.4%
Corporate governance 16.7% 3.5%

Qualitative factors
Controls 16.7% 3.5%
Financial reporting transparency 16.7% 3.5%
Risk positioning 30% 21% Credit risk concentration 16.7% 3.5%
70% Liquidity management 16.7% 3.5%
Market risk appetite 16.7% 3.5%
Licensing 10% 2.1%
Capital regulation 15% 3.2%
Regulatory environment 30% 21% Other prudential regulation 15% 3.2%
Supervision 30% 6.3%
Independence enforcement 30% 6.3%
Economic stability 33.3% 7.0%
Operating environment 30% 21% Integrity and corruption 33.3% 7.0%
Legal system 33.3% 7.0%
PPP% average RWA 50% 2.4%
Financial fundamentals

Profitability 15.75% 5%
Net income % average RWA 50% 2.4%
Liquidity 15.75% 5% (Market funds-Liquid assets) % total assets 100% 4.7%
Tier 1 ratio 50% 2.4%
30%

Capital adequacy 15.75% 5%

Shareholders equity % total assets 50% 2.4%
Efficiency 7% 2% Cost/income ratio 100% 2.1%
Problem loans % gross loans 50% 2.4%
Asset quality 15.75% 5%
Problem loans% (equity + LLR) 50% 2.4%
Lowest score 30% 9% Assigned to lowest scoring financial ratio 100% 9.0%
Why banks are different: JPD
• The mathematics of joint probabilities has been around for centuries, it’s
only recently been applied to default events
✓ probability of a bank defaulting is not standalone probability of default
✓ …but rather joint probability of he bank and its sovereign defaulting
simultaneously…
✓ …since the sovereign will (?) intervene and support bank, should it fail

• Joint probability of bank/sovereign defaulting is (in most cases) very, very

small (much smaller than the bank’s standalone PD)
• JPD can only improve ratings. At worst, they remain unchanged
• JPD does not only apply to sovereigns and banks – can be applied to
parent-and-subsidiary companies, joint ventures, etc., wherever some
form of external support that could prevent final default exists
Joint default probability
• No correlation between variables

-3
-2 3
-1 2
1
0
0
1
-1
2 -2
3 -3
Joint default probability
• Strong positive correlation between variables

-3
-2 3
-1 2
1
0
0
1
-1
2 -2
3 -3
Joint default probability
• Strong negative correlation between variables

-3
-2 3
-1 2
1
0
0
1
-1
2 -2
3 -3
Joint default probability

1
Asset B

-1

-2

-3

-4
Correlation = 0.0 Correlation = 0.75
-5
-5 -4 -3 -2 -1 0 1 2 3 4 5 -5 -4 -3 -2 -1 0 1 2 3 4 5

Asset A Asset A
Joint probability of default

Investment Bank

Bank fails 3%

Bank survives 97%

Joint probability of default

Sovereign

Sovereign Sovereign
fails survives

1% 99%
 Joint probability of default

Sovereign
Defaults Survives

0.4%
JPD
Fails
Investment Bank

2.6%
0.4% 2.6%
Survives

0.6% 96.4% 97%

1% 99%
 Joint probability of default

Sovereign
Survives
Defaults
No support Support

1.4%
JPD
Fails
Investment Bank

1% 1.6%
0.4% 1.6%
Survives

0.6% 96.4% 97%

1% 99%
Joint probability of default

𝐽𝑃𝐷 = 𝑃 𝐵 and 𝐺 = 𝑃 𝐵 ⋅ 𝑃 𝐺 + 𝜌𝐷 ⋅ 𝑃 𝐵 ⋅ 1 − 𝑃 𝐵 ⋅𝑃 𝐺 ⋅ 1−𝑃 𝐺

Default correlation
Link between asset
correlation and 2
default correlation
1.5

1
Asset correlation
0.5

30
2
𝐽𝑃𝐷 = 𝐹 Φ−1 𝑃𝐷𝐵 , Φ−1 𝑃𝐷𝐺 , 𝜌𝐴 -3 1
-2 0
-1 -1
0
1 -2
Copula (bivariate normal) 2 -3
3
 LOSS MEASUREMENT

Loss given Probability Exposure

default of default at default
EXTERNAL INTERNAL

Rating agencies KMV Logit Factor models

Internal ratings: LOGIT
• Scorecards – usually the start point
✓ Decide risk factors
✓ Obtain historical records
✓ Determine "strength" (relative importance) of risk factors
✓ These become the β coefficients in the equation:

z =  0 + 1 x1 +  2 x2 +  3 x3 +  +  n xn
Exposure Value of z
to some set when all Regression coefficients
of risk other risk describe size of contribution of
factors factors = 0 corresponding risk factor

✓ This z is then used in the logit regression function:

f (z ) = PD =
1
1 + e−z
Internal ratings: LOGIT
• Gives rise to the logistic regression predictor
• Example if death from heart disease
✓ Simple model uses only 3 risk factors (age, sex & cholesterol level) to predict
the 10-year risk of death from heart disease
β0 = − 5.0 (intercept)
β1 = + 2.0
β2 = − 1.0
β3 = + 1.2
x1 = age in decades – 5 (A)
x2 = sex where 0 = male and 1 = female
x3 = cholesterol level, in mmol/L – 5 (V)
1
Risk of death =
1 + e−z
where z = −5 + 2( A − 5) − 1(0 / 1) + 1.2(V − 5)
Internal ratings: LOGIT
100%

90% Age PD
10 0.00%
Proabability of dying in next 10y
80%
30 0.14%
70%
50 6.91%
60%
70 80.22%
50% 90 99.55%
40%

30%

20%

10%

0%
20 30 40 50 60 70 80 90

Age in years
 LOSS MEASUREMENT

Loss given Probability Exposure

default of default at default
EXTERNAL INTERNAL

Rating agencies KMV Logit Factor models

Internal ratings: KMV Link between loans and optionality
Merton (1974)

• Payoff on pure discount bank loan with face value B secured by firm's
asset value
▪ Firm owners repay loan if asset value (at loan maturity) > B (eg, A2)
▪ Bank receives full principal + interest payment
▪ If asset value A1 < B → default → bank receives assets
Payoff to bank lender

c.f. put option

on a stock

A1 B A2 Assets
Internal ratings: KMV Equity as a call option on a firm

• Value of put option on stock =𝑓(𝑆, 𝑋, 𝑟, 𝜎, 𝑇) where

• 𝑆 = stock price, 𝑋 = exercise price, 𝑟 = risk-free rate, 𝜎 = equity volatility, 𝑇
= time to maturity
• Value of default option on risky loan =𝑓(𝐴, 𝐵, 𝑟, 𝝈𝑨 , 𝑇) where
• 𝐴 = market value of assets, 𝐵 = face value of debt, 𝑟 = risk-free rate, 𝜎𝑨 =
asset volatility, 𝑇 = time to debt maturity
Value of equity (E)

B
0
A1 A2 Value of assets (A)
 Internal ratings: KMV
Distribution of asset
value at horizon
Value (assumed normal)

$100m

Asset value: 𝜎𝐴 = $10m

$90m Distance-to-Default = 2

Debt: $80m Default point

2.5% probability that normally distributed

assets fall by more than 2𝜎 from mean EDF

Today 1 yr Time
Internal ratings: portfolio credit risk
• 4 main credit risk models currently in circulation:
✓ Logit models
✓ Moody’s KMV model
✓ CreditRisk+ & CreditMetrics
✓ Factor models

• Operate in different ways and rely on slightly different data

KMV
✓ Uses a market pricing model

CreditRisk+ & CreditMetrics

✓ Transition matrices + discount credit curves

Factor models (such as Asymptotic Single Risk Factor)

✓ A kind of credit VaR model
 LOSS MEASUREMENT

Loss given Probability Exposure

default of default at default
EXTERNAL INTERNAL

Rating agencies KMV Logit Factor models

Internal ratings: factor models
• Credit risk is a combination of expected losses, unexpected losses and
catastrophic losses
• Expected losses covered by pricing and provisioning – these are
measured using 𝐸𝐴𝐷 × 𝐿𝐺𝐷 × 𝑃𝐷
• Unexpected losses are the focus of BII and the cause of much concern
in the banking world. Difficult to measure, but robust models (to
assign capital) are in use
• Catastrophic losses are, by definition, catastrophic and if capital were
to be allocated for these events, banks would have little left over to do
anything else. Insurance covers some of these, but much remains
unprotected
• To begin with, examine the components of expected losses: how are
they measured?
Internal ratings: factor models

Pricing and Basel II credit risk Unprotected: depends

provisioning capital equations on risk appetite

𝑝𝑑 𝑙𝑔𝑑 𝑒𝑎𝑑
Expected Unexpected Catastrophic
Frequency

losses losses losses

Losses
Similarities: market & credit risk

Unexpected
Market risk
loss

1% Expected
returns loss

-25% -20% -15% -10% -5% 0% 5% 10% 15% 20% 25%

Losses (%)
99% CI VaR
Similarities: market & credit risk

Unexpected loss
Credit risk

0.1%
returns Expected
loss

-25% -20% -15% -10% -5% 0% 5% 10% 15% 20% 25%

Losses (%)
99.9% CI CVaR
Similarities: market & credit risk

Market risk

Credit risk

-25% -20% -15% -10% -5% 0% 5% 10% 15% 20% 25%

Losses (%)
Internal ratings: factor models
• This model assumes a single factor drives the underlying process (whatever it
is)
• In describing the default process, it is commonly assumed that the single
factor is the state of the world’s economy and all loan values/obligor’s default
probabilities are tied to this in a "simple" way

Probability density
Internal ratings: factor models
• These are tied together by the correlation between them

Y =   X + 1 −   2

• X is an economic factor that influences all assets in portfolio (but to

different extents)
•  measures degree of ith asset’s exposure to systematic risk expressed by
X, and ε ~ N(0,1)
• Incidentally, this is the same model (and some of the same model
assumptions) made by Basel II in the now famous equations for calculating
regulatory capital
Internal ratings: factor models
• Monte Carlo (MC) methodologies used by rating agencies implicitly based
on simulation of default events
• Assumption: default occurs when value of an obligor’s assets falls below
that of its liabilities
• All assume that changes in obligor asset values are lognormally distributed
so that a normalized “distance” to default “threshold” (DDi) can be inferred
from default probability (PDi) associated with obligation’s credit rating: DDi
= N-1(PDi), where N-1(x) is inverse of standard normal distribution
Internal ratings: factor models

• In each simulation, correlated standard normal random variable is

drawn for each obligation in the portfolio, which is taken to represent
normalized change in obligor’s asset value over appropriate horizon
(ΔXi)
• If ΔXi < -DDi, a default is indicated and loss or recovery is drawn from
appropriate distribution
• Default losses accumulated for n assets in a pool of loans to give total
loss:
n
LT =  (X i  DDi ) Vi  (1 − RRi )
i =1

• where RRi is recovery rate and Vi is value of ith obligation

What is the Vasicek distribution?
• The Vasicek distribution is a distribution which describes losses due
to credit risk (much like the normal distribution describes losses due
to market risk )
• It is based upon a single factor risk model
• Loss distribution density is

f ( x; p ,  ) =
1−

 −1
exp ( 1− N −1
(x ) − N ( p ))
−1 2
+ (
1 −1
)
2
N ( x ) 
 2 2 
• The cumulative loss distribution is
 1 −   N −1 ( x ) − N −1 ( p ) 
F ( x; p ,  ) = N  
  
 
Basel II’s views on credit risk
• Basel II uses the Vasicek distribution in its description of credit losses
and the measurement of credit risk
• It assumes
✓ a single risk factor (usually GDP – or some other measure of economic health)
✓ all loans are tied to single risk factor by single correlation value
✓ longer maturity loans are more risky than short-term loans

✓ correlation varies with PD

✓ LGD not correlated with PD

What are the IRB equations?

 1 + (M − 2.5)  b 
  1
K = LGD   N 
  1 − 
−1
 −1



 N (PD ) +   N (0.999)  − PD   
  1 − 1.5  b 

 

 1 − e −50PD 
 = 0.24 − 0.12   −50 

 1− e 
 1 − e −50PD  b = (0.11852 − 0.05478  log(PD ))
2
 = 0.30 − 0.18   −50 

 1− e 
A closer look at the IRB equations
Unconditional Correlation between loan value
probability and state of the world


N
1
 1− 
−1
 −1

 N (PD ) +   N (0.999) 


 
Conditional State of the
probability world
• Conditional on the “state of the world” (0.999 = bad)
• Calculates the 99th percentile using one-factor model
A closer look at the IRB equations
• Example
• Assume PD = 0.04% and  = 0.41

1
=𝑁 × 𝑁 −1 0.04% + 0.41 × 𝑁 −1 0.999
1 − 0.41

1
=𝑁 × −3.35 + 0.41 × 3.09
0.59
= 𝑁 1.30 × −3.35 + 0.64 × 3.09

= 𝑁 −1.7889

= 3.68% “Stressed PD”