Risk Measurement and Management - PowerPoint PPT Presentation

About This Presentation
Title:

Risk Measurement and Management

Description:

J. K. Dietrich - FBE 525 - Fall, 2006. Risk Measurement and Management. Week ... Properties ... Interest rates stable, large post-war holdings of government bonds, ... – PowerPoint PPT presentation

Number of Views:66
Avg rating:3.0/5.0
Slides: 68
Provided by: jkdie
Category:

less

Transcript and Presenter's Notes

Title: Risk Measurement and Management


1
Risk Measurement and Management
  • Week 11 November 2, 2006

2
Measuring Holding-Period Risk
  • Price sensitivity of bonds is measured in terms
    of a bond price elasticity
  • This elasticity is called duration denoted d1,
    which is widely used by bond traders and analysts
    and is often available on quote sheets

3
Example of Duration
  • Assume a 10-year 8 coupon bond is priced at 12
    yield to maturity and has value of 77.4 and
    duration of 6.8
  • If yields changed immediately from 12 to 10,
    that is a 2/112 or 1.8 change in gross yield
  • The bond price should change about
    1.8 6.8 12.1

4
Duration as Time Measure
  • In 1930s, Macauley noted that maturity was not
    relevant measure of timing of payments of bonds
    and defined his own measure, duration
  • The definition of duration is (19-4)

5
Duration has two interpretations
  • Elasticity of bond prices with respect to changes
    in one plus the yield to maturity
  • Weighted average payment date of cash flows
    (coupon and interest) from bonds
  • Duration measure
  • Can be modified to be a yield elasticity by
    dividing by (1yield to maturity)
  • can be redefined using term structure of yields
    (Fisher-Weil duration noted d2)

6
Duration Calculations
  • Duration can be calculated for bonds
  • For level-payment loans (e.g. mortgages)

7
Duration is an Approximation
Derivative is used in calculating duration
Price (Par1.0)
Actual price change
Change predicted by duration
0
Yield to Maturity
8
Properties of Duration
  • Can be interpreted as price elasticity or
    weighted average payment period
  • Note when c0 that d1 M
  • When M is infinite d1 (1i)/i
  • Duration measure effect of parallel shift in
    interest rates
  • Other economic risks are not assessed

9
Duration as Risk Measure
  • Good
  • Balances reinvestment yield risk against capital
    gains risk
  • Widely used and clear mathematical expression
    assessing holding-period yield risk
  • Bad
  • Approximation and theoretical issues
  • Convexity adjustment only approximate improvement

10
Asset Liability ManagementDefinitions
  • Approach to balance sheet management including
    financing and balance sheet composition and use
    of off-balance sheet instruments
  • Assessment or measurement of balance sheet risk,
    especially to interest rate changes
  • Simulation of earnings performance of a portfolio
    or balance sheet under a variety of economic
    scenarios

11
History of ALM
  • After World War II to mid 1960s
  • ASSET MANAGEMENT
  • Interest rates stable, large post-war holdings of
    government bonds, deposit markets protected
  • Mid 1960s to late 1970s
  • LIABILITY MANAGEMENT
  • Interest rates rising, global financial markets
    developing (e.g. Eurodollars), regulation binding
    (maximum deposit interest rates)

12
History of ALM (continued)
  • Late 1970s to present
  • ASSET/LIABILITY MANAGEMENT
  • Use balance sheet composition or off-balance
    sheet instruments to management interest rate and
    other economic risks
  • Changing markets - increased competition from
    non-banks, foreign institutions
  • Goverment concerns - SL failures, Continental
    Bank and Texas banks, etc.

13
Measurement of Risk of Balance Sheet
  • Maturity gaps are common way to assess the
    sensitivity of a balance sheet to changes in
    interest rates
  • Assets and liabilities classified by maturity or
    repricing interval
  • Cumulative gap calculated
  • Not easy to interpret in terms of risk

14
Duration of Balance Sheet
  • Duration of a number of assets is
  • Duration of net worth in a portfolio is

15
Simulation
  • Computer simulation can handle more complex
    economic changes
  • Many simulations can assess sensitivity of
    earnings to changes
  • Regulators require and consultants can apply

16
Managing Interest Rate Risk
  • Change balance sheet composition
  • Adjust assets and liabilities until dE is at
    acceptable level
  • Use futures or options to adjust next exposure
  • What is source of value added?

17
Can Risk Management Add Value?
  • Return to risk-free portfolio is the risk-free
    rate
  • Investors can manage their own interest rate risk
  • Does risk distract management or prevent
    exploitation of competitive advantage?
  • Pleasing regulators and better understanding may
    be biggest advantage of ALM

18
Risk Management
  • Balance sheet management
  • ALM
  • Duration and immunization
  • Off balance sheet
  • Futures
  • Options
  • Swaps

19
Types of Derivative Contracts
  • Three basic types of contracts
  • Futures or forwards
  • Options
  • Swaps
  • Many basic underlying assets
  • Commodities
  • Currencies
  • Fixed incomes or residual claims

20
Managing Risk with Futures
  • Offset price or interest rate risk with contract
    which moves in opposite direction
  • Cross diagonally in the box
  • Identify contract with price or interest rate
    which moves as close as possible with the price
    or interest rate exposure
  • Imperfect correlation is basis risk
  • Not using futures or forwards can be speculation

21
Hedging
Bank Planning to Borrow
Insurance Hedge
Borrowing Hedge
Insurance Company with Premiums
22
Interest-Rate Options
  • Interest rates and asset values move in opposite
    directions
  • Long cash means short assets
  • Short cash means long (someone elses) asset
  • Basis risk comes from spreads between exposure
    and hedge instrument
  • Problem with production risk

23
Caps, floors, and collars
  • If a borrower has a loan commitment with a cap
    (maximum rate), this is the same as a put option
    on a note
  • If at the same time, a borrower commits to pay a
    floor or minimum rate, this is the same as
    writing a call
  • A collar is a cap and a floor

24
Collars Cap 6, floor 4
  • Profit

0
9400
9500
9600
Loss
25
Options and Product Pricing
  • Option pricing is well established technology
  • Black-Scholes approaches
  • Present value approaches
  • Simulation
  • In interest rates, lattice models used which are
    consistent with interest rate movements
  • Can model any cash flow with combinations of
    options

Rocket Science
26
Replication Futures with Options
Profit
Profit
Buy Call
Long
0
0
P0
P0
Loss
Loss
Write Put
27
Other option developments
  • Credit risk options
  • Casualty risk options
  • Requirements for developing an option
  • Interest
  • Calculable payoffs
  • Enforceable

28
Swaps
  • Exchange of future cash flows based on movement
    of some asset or price
  • Interest rates
  • Exchange rates
  • Commodity prices or other contingencies
  • Swaps are all over-the-counter contracts
  • Two contracting entities are called
    counter-parties
  • Financial institution can take both sides

29
Interest Rate SwapPlain vanilla, LIBOR_at_5.5
1/2 5 fixed
Company A (receive floating)
Company B (receive fixed)
2.5mm
2.75mm
1/2 6-month LIBOR
Notional Amount 100 mm
30
Issues in Hedging
  • Micro-hedging versus macro-hedging
  • Accounting
  • Regulation
  • Assumptions underlying hedging
  • Market liquidity
  • Covariance structure (second moments)
  • Notorious examples
  • PNC, IG Metall, Bankers Trust, Orange Cy,
    Long-Term Capital Mgmt (LTCM), BancOne

31
Overview of Credit Risk
  • Usual interpretation of credit risk is default on
    a loan or bond
  • New views of credit risk are focused on the
    change in the credit-worthiness of debt
    instruments as well as default
  • Risk changes will be reflected in the value of a
    portfolio over time as write-downs or downgrades
    short of default reduce value of claims
    (mark-to-market view of risk)

32
Default
  • Private debt (corporate and household) may not
    pay cash flows as promised
  • Late payments
  • Nonpayment of interest or principal
  • Other default or credit events
  • Violation of covenants and other creditor
    interventions in operations
  • Change in risk of default (e.g. highly leveraged
    transactions)

33
Credit Events
  • Probability of default (PD) can change affecting
    the value of default-risky securities
  • Upgrades and downgrades reflecting changes in PD
    are credit events
  • Recent progress has been made in quantifying
    these probabilities

34
Bond and Debt Ratings
  • Rating agencies
  • Standard and Poors (AAA to D)
  • Moodys (Aaa to C)
  • Fitch and Duff and Phelps
  • Migration of ratings, e.g. from BBB to BB
    (investment grade to below investment grade)
    represents credit risk
  • For example, change from BBB to BB has historical
    probability of 5.3 (SP, 1996)

35
Ratings and Defaults
36
Risk of Fixed Incomes
Probability
Maximum valueF
Future Value of Debt
37
Credit Losses
  • Three elements in credit losses
  • Estimated default probability (PD)
  • Loss given default (LGD)
  • Exposure at default (EAD)
  • Credit losses PDLGDEAD
  • Investors in debt securities will be concerned
    about all these elements in managing their risks

38
Credit Risk Analysis
  • Credit risk has become a major focus of rating
    agencies, regulators, and investors
  • Very important to capital market development
    (e.g. asset securitizations, loan syndications)
  • Enron, Global Crossing, and GE exemplify
    different stages of concern with these issues
  • Consulting industry in credit analysis
  • RiskMetrics (formerly J.P. Morgan)
  • KMV (academic based research)
  • Others (KPMG, PricewaterhouseCoopers, etc.)

39
Credit Risk Assessment
  • Default occurs when value of assets less than
    value of liabilities (insolvency)
  • Example of analysis used by KMV uses simplified
    estimates of variables
  • Must calculate market value of assets (market
    value of debt and equity) and variability of
    market value
  • Identify book value of liabilities

40
Motorola Debt and Equity
Total Market Value
41
Distance to Default Example
  • Motorola 2001-II (billions) Value of long-term
    debt 7.3 Book value of current
    liabilities 12.9 Total value of
    liabilities 20.2 Market value of assets
    56.6 Standard deviation of change in market
    value 16.4
  • Market value ? standard deviation of percent
    change 9.3 billion

42
Reduced Probability of Default?
  • Estimated default point in example is midway
    between book value of current liabilities and
    long-term debt
  • Theory is that long-term debt does not require
    immediate payment, short-term liabilities may
    allow some flexibility
  • KMV uses historical data to fine-tune this
    estimate

43
Estimated Distance to Default
Market value to default point 40.0
56.6
20.2
12.9
TMV
CLLTD
CL
Default point (estimated as midpoint) 16.6
44
Distance to Default 12-31-01
  • Total Value of Assets (from Capital Structure
    and Financial Statements) E LTD CL
    TA33.9 8.1 9.7 51.7
  • Book value of LTD and CL 8.4 and 9.7Midpoint
    estimate of default point 13.9Std Dev 16.4
    51.7 8.48

45
Probability of Default
  • KMV has used historical data to relate distance
    from default to probability of default
  • That measure is proprietary (not available)
  • As example, Motorola is rated A3 by Standard and
    Poors, historically associated with a default
    rate of about .82 over next five years (.61 in
    Moodys experience)

46
Private Firm Default Risk
  • KMV estimates non-traded firm risk by using
    market-traded comparables
  • Data base on 35,000 traded firms globally
  • Valuations of private firms and risk estimated by
    using EBITDA/Assets ratios
  • KMV estimates default probabilities for private
    firms based on data on 300,000 firms in 30
    countries
  • Estimates depend on EBITDA??0

47
Credit Risk in Portfolios
  • Individual assets have probability of default and
    risk and discussed last week
  • Loans in portfolios will have an interdependent
    risk structure due to correlations in defaults
  • Credit risk within portfolio context is a major
    advance in credit risk management
  • Search for a summary measure of portfolio risk
    led to the concept of value at risk

48
Value at Risk (VAR)
  • Value at risk (VAR) looks at risk of portfolio
    accounting for covariance of assets
  • Risk is defined in terms of likelihood of losses

49
VAR and Capital
B
Capital
50
Portfolio Credit Risk
  • Credit risk different than usual portfolio risk
    analysis
  • Returns are not symmetric
  • Concentrations of exposure complicate losses
  • Major issue is correlation of defaults and losses
    given default
  • We will discuss approach followed by
    CreditMetrics
  • Other approaches exist (including KMV)

51
Credit Risk as Rating Changes
  • Increased credit risk Default
    CCC B BB
  • Same credit risk (BBB) BBB
  • A AA
  • AAA
  • Less credit risk

52
Rating Migrations (BBB rating)
Source Standard Poors
53
Two Bond Rating Migrations
54
Probability of Default Two Firms
Value of Firm B
Probability 1/2
Probability 1/10
Probability 1/100
Default Point B
Value of Firm A
Default Point A
55
Loss Given Default
56
Simplified Road Map
Compute exposure profile Of each asset
Compute the volatility Of value caused by Up
(down)grades and defaults
Compute correlations
Portfolio value-at-risk due to credit
Source Introduction to CreditMetrics (1997)
57
Required Resources
  • Default probabilities (or ratings)
  • Migration probabilities
  • Historical data requirements
  • Approaches to estimating correlations
  • Complete data on types of credits and estimations
    of losses given defaults
  • Exposures to classes of risks
  • Models and simulations of value changes given
    credit events

58
Credit Portfolio Risk
One Asset
Many Assets
Frequency
Frequency
0
0
Return
Return
59
Incremental Risk
  • Introduction to CreditMetrics provides good
    examples (in Section 5)
  • Importance portfolio risk is the marginal risk
  • Marginal risk considers portfoliorisk
    implications

10
High risk and large size
?
10mm
Credit Exposure
60
Example Portfolio
Source Creditmetrics Technical Document (April
2, 1997)
61
Credit Risk Management
  • Derivatives Single-name v. multi-name
  • Types of credit derivatives
  • Total return swap
  • Credit risk swap
  • Credit risk option
  • Credit inter-mediation swap
  • Credit spread derivative
  • Default substitution swap
  • Over 400 billion notional amount 2000-IV

62
Hedging Credit Risk
Hedging Instrument Payoff
Change in Portfolio Value
0
Risky Outcomes
63
Example of Total Return Swap
  • 3-year 8 coupon bond
  • If default probability increases from 10 to 20,
    bond return is 8 - 6.4537 1.5463 (coupon
    minus loss due to downgrade)

64
Total Return Swap
(8-6.5437)
Company A (pay total return)
Company B (pay fixed)
154,626
750,000
7.5
Notional Amount 10 mm
65
Total Return Swap
  • Difference between payments and receipts by total
    return receiver is compensation for risk
  • Total return payer receives cash in case of
    downgrade as in example, subsidizing loss
    realized on balance sheet
  • Can have other swap types, as in default swap

66
Limitations of Derivatives
  • Market limited to single name and portfolio
    instruments
  • Typically individual corporate borrowers
  • Some portfolio of commercial loans
  • Market not developed for consumer credit
  • Growth of consumer market but most
    risk-management consists of sales of loans
  • Global market ready for consumer-risk derivatives

67
For Next Classes
  • Prepare First American Bank Credit Default Risk
    case for November 9
  • Read Chapters 23 and 24 for discussion in class
    on November 9 and November 16
  • Read KMV paper and Creditmetrics paper before
    November 16 class
  • Teams should schedule appointments with me
Write a Comment
User Comments (0)
About PowerShow.com