TOPIC 1

Market Portfolio: all traded stocks with weights given by the ratio of each company’s market value and the total market value. If investors have mean-variance preferences, they invest in a
combination of a riskless asset and the tangency portfolio. Hence, tangency portfolio represents the demand for the total assets in the economy. The market portfolio represents the
supply of all assets. In equilibrium the prices adjust in such a way that demand equals supply. Hence, the market portfolio should be the tangency portfolio.
Vi
Formula: wiM =
V1 + . . . + VN
If PSG = 1 =>D G↓ P G↓ E (rG )↑ => D G↑ => non sense
Tangency Portfolio: represents the demand for the total assets in the economy. At equilibrium they’re equal
S = stock and G = gold
VARIANCE: σ p2 = wa2 σ a2 + w 2 σ 2 + 2wa wb σa σ b ρ
b b
ρ = −1 σp = | wA σA − wB σB | don’t use it directly but derive it from the one above. It’s basically: (a + b ) 2 = a 2 + 2 a b + b 2

PEARSON COEFFICIENT: ρ = 1 no bene t to diversi cation, ρ = − 1 the volatility is 0

1
UTILITY FUNCTION: U = E (r ) − v a r (r )
2τ
EXPECTED RETURN: E (r ) = wa ra + wb r b
CAPM: E (Ri ) − rf = β (E (rM ) − rf )
C O V (ri , rM )
β COEFFICIENT: β = measure of an asset’s systematic risk, which is the risk that cannot be diversi ed away. It represents sensitivity of the asset’s return to the return of the
σ 2 (rM )
market’s portfolio. For a portfolio P* the measure of the risk is the σ 2 , while for an asset is β
P* βP = w ⋅ βM + (1 − w )β 0
The higher an asset co-moves with the market, the higher COV, the riskier because ↑β
Risk free: β =0 asset has only the individual risk that can be diversi ed away. same return as risk free rate: no need for a risk premium. We cannot bene t from diversi cation u = rf

Market: β =1 its price tend to move linearly to the market. ri = rm

Riskier: β >0 E (r ) > rf asset ↑ the riskiness of market portfolio because the asset ↑ the market risk of the portfolio. Investors require higher return to compensate for it.

More Secure: β <0 E (r ) < rf asset ↓ the riskiness of market portfolio accepting lower return than the risk-free rate. You buy it as an insurance for the portfolio.

α COEFFICIENT: α = E (Ri ) − [rf + β (E (rM ) − rf )] is the expression of the abnormal return. You can use it to judge someone’s portfolio.

E (Ri ) = rf + β (E (rM ) − rf )
α >0 UNDERPRICED asset asset gained a higher payo than expected from CAPM: its return is > than the systematic risk
α <0 OVERPRICED asset asset gained a lower payo than expected from CAPM: its return is < than the systematic risk

EFFICIENT FRONTIER: describes the portfolios that for a given expected returns μ have the lowest possible σ 2.

P = (rf , P*)

P = (1,0) in nitely risk averse

P = (0,1) less risk averse
P = (−1,2) less less risk averse (NEVER RISK LOVER!)

E (rP ) − rf
SHARPE RATIO: SR = how much reward you get for any unit of σ P. P* is the highest SR portfolio (slope of CML)
σP
OWNING CORP BOND ∼ WRITE PUT
F
VALUE of a FORWARD contract: V =S− The bondholders have e ectively made a loan that requires repayment of B dollars, where B is the face value of bonds. If, however, the value of
(1 + r )T the rm (V) is less than B, the loan is satis ed by the bondholders taking over the rm. In this way, the bondholders are forced to “pay” B (in the
sense that the loan is cancelled) in return for an asset worth only V. It is as if the bondholders wrote a put on an asset worth V with exercise price B.

FORWARD price from V=0: F = S (1 + rf )T Alternatively, one can view the bondholders as giving equity holders the right to reclaim the rm by paying o the B dollar debt. That is, as if the
bondholders have issued a call to the equity holders.

TOPIC 2
HEDGING IN EQUITY MARKETS
PAYOFF of a CALL: max(ST − K ,0)
PAYOFF of a PUT: max(K − ST ,0)
Upper bounds: c ≤ cA ≤ S
p ≤ K B (t , T ) and PA ≤ K
Lower bounds: c > max(S − K B (t , T ),0)
p > max(K B (t , T ) − S,0) and P ≥ max(K − S,0)
K
PUT-CALL PARITY: p + s0 = c + if violated ⟹ arbitrage.
(1 + rf ) t
To avoid arbitrage: d < 1 + rf < u
D e + P1e − P0 b u l l l (k 1, k 2 ) + b e a r (k 1, k 2) = b o n d
E. RETURN STOCKS: E (r ) =
P0
Cu − Cd δP ΔP
DELTA (∆): Δ= = = sensitivity of the option’s price to changes in the price of the underlying security. Number of units of stocks in the hedging portfolio
(u − d )S δS ΔS
u Cd + d Cu
B: B =
(u − d ) ⋅ 100
100
PRICE of a CALL : C =Δ⋅S +B ⋅
1 + rf
q Cu + (1 − q )Cd E*(C1) 1+r −d
C = = with q which is RISK NEUTRAL PROB: 0 < q = <1
1+u 1 + rf u −d
STRADDLE Price: Pstraddle = C + P
STRADDLE BEP: B E P = Pstraddle ⋅ (1 + r )t
fi
fi
ff
fi
fi
fi
fi
fi
ff
fi
ff
ff
fi
fi
fi
 δc δc 1 δ2c 1 δ2c 2 1 δ2c
TAYLOR EXPANSION: Δc = ΔS + Δt + ΔS 2 + Δt + ΔS Δt + . . .
δS δt 2 δS 2 2 δ t2 2 δtδS
BLACK-SCHOLES CALL: c = S (d1) − K e −r(T−t) (d 2 )
S 1 ↑K ↓C
ln( ) + (rf + σ 2 )(T − t )
K 2 ↑K ↑P
d1 =
σ T −t

d 2 = d1 − σ T − t
BLACK-SCHOLES PUT: p = K e −r(T−t) (−d 2 ) − S (−d1) Remember: (−d1) = 1 − (d1)
δP
DELTA OF A CALL: Δ= = (d1) (for European options on non-dividend paying stocks). The higher the P, the lower the Δ B E Pcall = C + K
δS
δP
DELTA OF A PUT: Δ= = (d1) − 1 (for European options on non-dividend paying stocks). The higher the P, the lower the Δ B E Pput = P + K
δS
δΔ δ2 P
GAMMA: Γ= = sensitivity of the option’s delta to changes in the price of the underlying security
δS δ S2
if it’s too high we have to rebalance the portfolio. If it’s too low, hedging is more stable so we don’t need to rebalance too much.
δP
VEGA: V = sensitivity of the option’s price to changes in the volatility of the underlying security
δσ
For EUROPEAN CALL: V = S (T − t ) (d1) (on non dividend paying stocks)

RHO: ρ sensitivity of the option’s price to changes in the diskless interest rate
δP
THETA: θ =
δt
δP δP 1 δ2 P 1 DELTA HEDGING
For a delta neutral portfolio: ΔP = Δt + ΔS + ΔS 2 = θΔt + Δ ⋅ ΔS + ΓΔs 2 Long Call: Δ ca ll > 0 *SHORT ∆ shares
δt δS 2 δ S2 2
Long Put: Δ pu t < 0 *BUY ∆ shares
For a CALL: the higher the σ, the higher the P AND the lower the σ, the lower the P
Short Call: Δ ca ll > 0 *BUY ∆ shares
For a PUT: the higher the σ, the lower the P AND the lower the σ, the higher the P
σ Short Put: Δ pu t < 0 *SHORT ∆ shares
WEEKLY VOLATILITY: If it’s replicating: do viceversa *
52
HEDGING IN FIXED INCOME MARKETS
T Co u p o n Principal T ct T
∑ (1 + y) t ∑ ∑
BOND PRICE: P = + = = δt ct y = Y T M rate that equalizes the PV of the coupons and the principal to the market price of the bond
(1 + y)T (1 + rt ) t
t=1 t=1 t=1
Long-Term Bonds: These bonds have a longer time to maturity.
1 • A small change in interest rates can signi cantly impact the PV of their future
SPOT RATE: δt = rt is the yield on a maturity t zero coupon bond with annual compounding. CFs. This translates to a larger price change.
(1 + rt ) t • However, since the YTM incorporates both coupon payments and capital
gains/losses, the YTM uctuation might be smaller for long-term bonds.

(1 + r 2 ) 2 Short-Term Bonds: Conversely:

FORWARD RATE: (1 + r 2 ) 2 = (1 + r1)(1 + f1,2 ) ⟹ 1 + f1,2 = • a change in interest rates has a less pronounced e ect on the PV of their CFs
1 + r1 due to their shorter maturity. This translates to a smaller price change.
• However, the YTM of short-term bonds is more sensitive to interest rate

(1 + rs ) s (1 + fs,t )t−s = (1 + rt )t
changes because the coupon payments are a smaller portion of the total
More generally: return. Any change in interest rates will have a more signi cant impact on the
YTM uctuation for a short-term bond.
e
$P1,t+1
ARBITRAGE CONDITION: 1 + i1,t = return per $ of holding a 1y bond for 1y = expected return per $ of holding a 2y bond for 1y
$P2,t
P1,0 + P2,0 = P2,c return of holding a 1y bond for 1y and a 2y bond for 2y (both 0 coupons) = return of holding for 2y a bond with coupons

(P1 − P0 ) + C1 = (P e − P0 ) + C2 + (P e − P0 ) + C3 R of a 1y bond held for 1y with c= E(r) of a 2y bond held for 1y with c = E(r) of a 3y bond held for 1y with c
1,t+1 2,t+1
1 N 1 dP
t c e −rti = −
P ∑ i i
DURATION: D = weighted average time to receive the bond's cash ows.
P dr
i=1
Time to maturity: the longer, the higher the duration
Coupon rate: the larger the coupon, the lower the duration

ΔP
≈ − DΔr R (bond ) = This adjusts D m ac to account for r changes. It’s commonly used to assess interest rate sensitivity.
P
1 N
t c e −rti
P ∑ i i
MACAULAY DURATION: DMac =
i=1
1 N ti ci D
Dmod for annual compounding: Dmod = mac
P ∑ (1 + R )ti
DMac for annual compounding: DMac =
(1 + R )
i=1
ΔP
if r↓ by x% you use these formulae to calculate how much the bond price will ↑: in % terms: ≈ − Dmod ΔR and in $ terms: ΔP ≈ − Dmod Δr P
P
Duration as a n. immunization: DA ⋅ Assets = DL ⋅ L iabilities
x1 P DMod,2
Using this hedge ratio: =− 2
x2 P1 DMod,1

1 T t (t + 1)ct
P ∑ (1 + R ) t+2
CONVEXITY with annual compounding: C = it takes into account non-linear functions and improves approximation.
t =1
ΔP 1
≈ − D Δr + C (Δr ) 2
P 2
RV < I V => replication is cheaper than buying CALL/PUT
Dividend RV >I V => buy call/put is cheaper than replicating
PERPETUITY: P =
r R V= real volatility
Discount rate
I V = implied volatility
1+i e
Uncovered Interest Parity (UIP): Et st+1 − st = rt − r*
t E = E
1 + i*
𝒩
𝒩
𝒩
𝒩
𝒩
𝒩
𝒩
𝒩
𝒩
fl
fi
fl
fi
ff
fi
fl
 TOPIC 4: VALUE AT RISK • Reality: The BS model's assumptions are often violated in practice. The IVs • Assumptions Limitations: Works primarily for zero-coupon bonds (ZCB) and
differ across various dates, strikes, and maturities, indicating that market assumes a constant risk-free rate.
RISK MEASURE: numbers that capture risk measure, but there are two problems: behavior deviates from the BS model. The IV is not constant. • Merton model works just with single maturity debt, while KMV allows multiple
underlying distribution unknown and cannot measure risk directly. The typical one.
Time Value Focus: Since intrinsic value is model-independent, modelers should
ones are: σ - VaR - E S Example:
focus on the time value of options. IV is calculated with the BS model, yet its
Investors that take into account mean-variance preferences, σ can be sufficient, widespread use doesn't imply agreement with BS assumptions. • Merton Model: Models a firm's equity as a call option on its assets, with
unfortunately that’s not true. default occurring when asset value falls below debt value.
Advantages of Using IV even if it's conceptually wrong:
VaR: is a quantile on the distribution of P/L. The loss on a trading portfolio such REDUCED form models
• Easier to Express Views: more convenient for expressing market views
that there is a probability p of losses equalling or exceeding VaR in a given compared to quoting option prices. PROS:
trading period and a (1 − p ) probability of losses being lower than VaR. It • Link to Market Prices: Can map observed credit spreads to default
summarizes the worst loss over a target horizon that won’t be exceeded with a • Unit-less Nature: Unlike option prices, IV is unit-less, making it easier to probabilities.
compare.
given level of confidence • Basis for Credit Derivatives: Forms the basis for most credit derivatives
• Intrinsic Value Independence: IV does not depend on the intrinsic value of pricing models.
-VaR(p)
∫ −∞
the option. • Flexibility: Can be adapted to various market conditions and credit
p = P [q ≤ − VaR(p)] = f (q ) d q instruments.
• Normal Return Distribution Benchmark: The BS model provides a
benchmark; deviations in IV reveal deviations from normality. CONS:
VaR N d a ys = VaR1 d a y ⋅ N only correct with iid normal data Interpreting IV Patterns: • Exogenous Defaults: Default is modeled using an exogenous intensity
• OTM vs. ATM Options: Higher IVs for OTM options compared to ATM options process, which may not capture all firm-specific risks.
| VaR ( x %) | = σ * 𝒩 −1( x %) − μ Π suggest fat tails in the return distribution. • Data Requirement: Requires extensive market data to calibrate the model
• Left Tail Heaviness: A higher IV for OTM puts than for OTM calls indicates a accurately.
The sign of μ is negative in the formula above because: heavier left tail in the return distribution. • Simplifying Assumptions: Assumes certain market conditions which may not
If μ is high it’s a positive BLACK SCHOLES METHOD hold true in all scenarios.
Example:
VaR of a Bond:
• If σ ↑, c↑ and p↑ • Credit Derivatives Pricing: Used in the valuation of credit default swaps (CDS)
σ b ≈ | D m o d | σr and other credit derivatives.
• ATM option prices are approximately linear in σ
BONDS: are based on rating models and the most important rating agencies are
VaR b on d ≈ Φ−1( p ) ⋅ | D m o d | ⋅ σ r ⋅ Π = | D m o d | ⋅ VaR yiel d • σ is the only one free parameter in the BS model S&P or Moody’s
−
It’s an approximation due to assumptions: • I V = σ = f 1(c ) • Debtors are considered in default as soon as they miss a payment obligation
REALIZED VOLATILITY: on any coupon or principal.
• Linearity
• Parallel shifts in the yield curve
uses intraday data to calculate volatility. Purely data driven. Not always easy to • Probability of default for 2 periods of a B bond:
get good, clean intraday data for with long time series. Can use it for volatility 2 =P P
PBD
Furthermore we need to know the VaR (and σ) for the yield forecasts: HAR-RV (heterogenous autoregressive RV model). B A A D + PBB PBD + PBD PDD
• Easy to obtain for standard maturities R Vt +1,1d = α + βD R Vt,1d + β W R Vt,1w + βM R Vt,1m + ϵt +1,1d (( μ− σ )t +σ tZ )
Vt = V0 ⋅ e 2
• To calculate for arbitrary maturities we need to convert to standard maturities where R Vt,1d is the daily RV calculated using 5 minute returns.
Moreover, VaR is meaningless without p and holding period. The var requires the V 2
shape of the distribution.
CBOE VOLATILITY INDEX (VIX): ln( 0 ) + ( μ − σ )T
F 2
real-time index that represents the market’s expectations for the relative Z <−
Under normal assumptions VaR ( p ) = Π ⋅ 𝒩 −1( p ) ⋅ σ σ T
strength of near-term price changes of the S&P 500 Index (SPX). It’s the fear
Π= portfolio value index. MERTON MODEL:
COHERENT RISK MEASURE: must satisfy these 4 properties When the VIX is very high: Et (r M ) is very high so I should ↑ my exposure to Consider a firm with risky assets with a market value of V at time t.
• Monotonicity: X ≤ Y ⟹ φ ( X ) ≥ φ (Y ) the market. The firm is financed by debt and equity and the firm's outstanding debt is a
When the VIX is very low Et (r M ) is very low so I should buy the VIX index. single ZCB with Face value F, maturing at date T. The market value of the debt at
• Subadditivity: φ ( X + Y ) ≤ φ (Y ) + φ ( X )
The GARCH is the model that seems to be the most responsive and therefore t is B t. The market value of the firm's equity at t is St.
• Positive homogeneity: φ (c X ) = c φ ( X ) with c > 0
assimilates recent data the fastest. The HS seems to be the least responsive. The distance is measured in terms of the number of standard deviations away
• Translation invariance: φ ( X + c ) = φ ( X ) − c HOW TO CALCULATE VaR: from the default point.

Monotonicity Subadditivity Positive Translation • Non parametric method: historical simulation: no statistical models assumed V <F as T − t → 0 chance of default ↑ (or d↓)
homogeneity invariance
- Univariate HS V >F as T − t → 0 chance of default ↓ (or d↑)
Volatility NO YES YES NO
• Sort all historical returns from smallest to largest. Distance to default: d ↓ if σ ↑ or F↑ and d↑ if μ t↑ or Vt↑
YES. Violated YES. Violated if
VaR YES YES th
only for fat tails liquidity risk
• VaR(p) is p % smallest value V 2
ES YES YES YES YES - ln( t ) + ( μ − σ )(T − t )
Multivariate HS F 2
d =
Expected Shortfall (ES) is the expected loss conditional on VaR being violated. • Calculate historical portfolio returns and proceed as above. σ T −t
• ES takes the shape of the tail distribution into account while VaR does not, ES is
- ES estimation with HS:
Prob(Vt < F) = Φ(-d) or N(-d)
usually more complicated to compute as it requires to know the shape of the • Use HS to get VaR
distribution in the tail. Given that events in the tail are rare, it is usually hard to Pricing: B 0 + P0 = F e −r T
estimate. • Estimate ES by the mean of all observations ≥ VaR
- Importance of window size A bondholder can hedge the credit risk on the bond by buying a
• ES is a coherent risk measure, complement to VaR.
European put option on the asset value of the firm with strike price F and time to
• Useful especially for option positions, fat tails • Tradeoff: sensitivity vs structural changes expiration T denoting the price of the put at T=0 P0 and the value of the risky
• Sample size: minimum 3/p Bond at T=0 B0
∫ -VaR(p) q f (q ) d q 1 -VaR(p)
p ∫ −∞
E S = E [ loss|loss ≥ VaR(p)] = − −∞ =− q f (q ) d q • Parametric methods: based on estimating distribution of returns to get risk
∫ -VaR(p) f (q ) d q forecast. Start with estimate of covariance matrix. Use distribution Price of a put: P0 = F e −r T 𝒩(− d 2 ) − V0 𝒩(− d1 )
−∞
assumptions for residuals S
VOLATILITY MODELS: ln( ) + (rf + 1 σ 2 )(T − t )
Backtesting K 2
Pt d1 =
r t ∼ 𝒩(0, h t ) but Pt = − 1 is bounded from below so we must use ln Backtesting is a method that uses historical data to test an investing or trading σ T −t
Pt−1 strategy to determine whether it would have produced returns over a specific
returns to let it go until −∞ . We know that ln(1 + x ) ≈ x so period of time. d 2 = d1 − σ T −t
Pt Stress testing
R t ≈ ln(1 + R t ) = ln( ) = rt 1 B 1 F e −r T − P0
Pt−1 It’s goal is to create artificial outcomes and see how models cope in order to YTM: y T = − ln 0 = − ln
asses the stability of financial institutions and to check their ability to cope with T F T F
σ 2 (t ) = h (t ) = f (r (1), . . . , r (t ) different scenarios. DEFAULT SPREAD: sT = yT − r
Profit
σ2 = σ2 RAROC RAROC = 1
s T = − ln(Φ(d 2 ) +
V0
Φ(− d1 ))
1d VaR T F e −r T
σ 2 = σ 2 ( R1 + R 2 + R3 + R 4 + R 5 ) = 5σ 2 + 4 ⋅ 2 ρ σ 2 3 ⋅ 2 ρ 2 σ 2 2 ⋅ 2 ρ 3 σ 2 1 ⋅ 2 ρ 4 σ 2 VaR for a portfolio can be interpreted as the prudent level of capital needed to
5d
hold the portfolio. Then RAROC is the rate of return on capital. low leverage: s T ↑(increases) with maturity
VaR (95%)1d = σ 1d ⋅ 𝒩 −1(95%) TOPIC 5: CREDIT RISK 2 components: high leverage: s T ↓ with maturity
VaR (95%) 5d = σ 5d ⋅ 𝒩 −1(95%) • Risk of default: measured by probability of default (PD) over a given horizon Limitations of the Merton model: Asset value: When both equity and debt of
• Size of loss if default occurs (loss given default, LGD) or, equivalently, recovery the firm are traded , asset value can be obtained by. summing the market values
r (t ) rate conditional on default of equity and debt. In practice, many loans are not traded and often we only
z (t ) = rt = ht zt
h (t ) observe market price for Equity. + it assumes constant riskless rate of interest +
Challenges related to credit risk:
single debt maturity and no distinction between junior and senior debt.
1 W−1 2 • Asymmetry: limited upside but substantial downside Extensions to the Merton model: Black and Cox (Models in which default can
W ∑ t−i
MA model: h t +1 = r • Credit returns are skewed and fat-tailed occur at any time, rather than just at the maturity date, if the asset hits a
i=1 • Defaults are rare: difficult to estimate default probabilities or default boundary K); Kim, Ramaswamy and Sundaresan (Stochastic interest rates);
EWMA model: Thanks to the exponentially weighted average, it’s possible to correlations from past data. However, measures of credit risk on portfolios of Anderson and Sundaresan (Bargaining between debtholders and management
estimate the volatility using past returns and past volatility. Moreover, you have defaultable securities are extremely sensitive to these correlations. in the face of costly liquidation); Collin-Dufresne and Goldstein (The firm's option
weight to choose how much you want to take into account returns or volatility. to issue additional equal-priority debt in the future can increase credit spreads of
• Greater reliance on models than in the context of market risk previously issued debt.
EWMA is better because recent observation should be more important than the
past. • Mark-to-market changes (prior to default) because of changed expectations KMV METHODOLOGY: empirical observation that default tends to occur when
regarding default risk or because risk-free term structure changes.
asset value is somewhere between short term debt and long term liabilities. No
h t = (1 − λ )r 2 + λ h t−1 • transitions probabilities: The difference is due to the fact that transitions
t−1 longer assume prob. of default is given by Φ(-d*) but instead estimate default
probabilities are actually time-varying. prob for a given dist to def d* and a given time horizon directly using a large
GARCH model: the best method to use, you are more flexible: α + β ≥ 1
CREDIT RATINGS models dataset on firm defaults. These estimated prob are called Expectd default
h T+1 = ω + α r T2 + β h T frequencies EDF.
PROS:
STOCHASTIC VOLATILITY σ (r t , t ) : stochastic means that some variable is Vt 2
• Ease of Use: Simple and straightforward to implement. ln( ) + ( μ − σ )(T − t )
randomly determined and cannot be predicted precisely. However, a probability • Widely Accepted: Commonly used and understood in financial markets. d* = F* 2
distribution can be ascertained instead. In the context of financial CONS: σ T −t
modeling, stochastic modeling iterates with successive values of a random
• Averaging Across Firms: Assumes all firms within a category are
variable that are non-independent from one another. What non-independent homogeneous, which isn't accurate. 1
means is that while the value of the variable will change randomly, its starting F * = short term debt + long term debt
• No Adjustment for Systematic Risk: Typically do not adjust for variations in 2
point will be dependent on its previous value, which was hence dependent on credit risk over the business cycle.
its value prior to that, and so on; this describes a so-called random walk. RECOVERY RATE: financial term that refers to the percentage of a loan or debt
• Non-Normal Distribution: The empirical distribution of changes in portfolio that is recovered by creditors when a borrower defaults. In other words, it
Examples of stochastic models include the GARCH model used in analyzing value is non-normal and asymmetric.
time -series data where the variance error is believed to be represents the portion of an investment that creditors can expect to get back
• Business Cycle Averaging: More defaults occur during recessions than after a borrower fails to meet their debt obligations, such as in the case of
serially autocorrelated. Many fundamental options pricing models such as Black booms, but the model averages over the cycle.
Scholes assumes constant volatility, which creates inefficiencies and errors in bankruptcy.
Example:
pricing.
• Usage in Financial Institutions: Banks using credit ratings from agencies like • Risk Assessment: Investors use the recovery rate to assess the risk associated
with investing in bonds or other debt instruments. A higher recovery rate
d r t = μ (r t , t ) d t + σ (r t , t ) d Wt − d Z t S&P, Moody's, and Fitch to assess the creditworthiness of bond issuers.
implies lower potential losses in the event of default.
IMPLIED VOLATILITY (IV): STRUCTURAL form models
• Pricing of Debt Instruments: Debt instruments are priced based on the
a metric capturing the market's view on the likelihood of future changes in a PROS: likelihood of default and the expected recovery rate. Higher recovery rates can
given security's price. Used to project future price movements and supply and • Fundamentals-Based: Directly relate probabilities and recovery rates to firm justify lower interest rates.
demand dynamics. Essential for pricing options contracts, as it reflects the fundamentals.
• Portfolio Analysis: Financial institutions and institutional investors use
market's expectations of future volatility. • Adjusts for Default Correlations: Can adjust for higher default correlations recovery rates to manage risk and diversify their investment portfolios.
Since IV is not directly observable, it is calculated using options pricing models within the same industry or region.
• Captures Firm-Specific Risk: Allows for detailed analysis based on individual • The recovery rate condition depends on the seniority of then instrument held
like Black-Scholes (BS).
firm characteristics. • There are different kinds of recovery rates based on the seniority class. The
Theoretical vs. Real-World Behavior: CONS: hierarchy is the following:
• Black-Scholes Assumptions: In an ideal world, the volatility (σ) would be • Complexity: More complex to implement and require detailed firm-specific - First lien
constant across all dates, strikes, and maturities. Under these conditions, IV data.
- Senior secured
would match the security's return volatility. • Data Availability: Difficult to find accurate market value of the firm's assets
(V). - Senior unsecured
 - Senior subordinate For each mortgage in the pool we can compute the time to maturity and the • Beginning October 2008, the federal reserve banks pays interest on required
Credit VaR: coupon, and then the WAM and the WAC are simply the weighted averages of reserve balances and excess balances
time and coupon, where the weights are relative to the size of each mortgage. FEDERAL FUNDS RATE
• The empirical distribution of changes in the portfolio value is non-normal. It is The speed of prepayment is more complicated. The industry practice is to refer • Interest rate at which depository institutions lend balances at the Federal
asymmetric with fat left tail. Assuming normality for the purpose of VaR to some average prepayment rates to describe the speed of prepayment. It is Reserve to other depository institutions overnight.
calculations would be inappropriate in this context customary to describe the value of a MBS in terms of its spread of prepayment. • When Fed is selling securities it is taking cash out of the system and draining
• Issues: Transitions are not independent for different bonds; Transition matrix 2 most common measures of prepayment are: reserves.
should depend on the business cycle, Recovery rates also vary across business • CMM: this measure assumes there is a constant probability that the average • Draining reserves increases the fed funds rate at which banks with deficit
cycle (lower in recessions) mortgage will be prepaid after the next coupon. reserves may be able to borrow.
TOPIC 6: CREDIT DERIVATIVES AND ASSET BACKED SECURITIES • PSA experience: the 100% PSA established by the Public Securities • Borrowing and lending in fed funds market can be done directly by the banks
CREDIT DEFAULT SWAPS (CDS) Association makes the following assumptions: or through brokers.
- CRP-0.2% of the principal is paid in the first month • Transactions typically take place overnight, lending and borrowing is
It’s the most popular form of credit derivative. Protection buyer receives a
- CPR increases by 0.2% in each of the following 30 months and unsecured.
payoff if a certain credit event related to a reference entity occurs.
- CPR then levels off at a constant annual prepayment rate of 6% until • Effective Fed funds rate is the the volume weighted Fed funds rate at which
If the event occurs, the protection buyer has the right to sell the bonds at face
value or receives a contingent amount, often specified as the difference between maturity. reserves are lent and borrowed.
the face and market value of the bond at the time of the event. The annualized conditional prepayment rate CPR: this rate measures the • Cutting the target rate is viewed as easing credit and increasing target is
In exchange, the protection buyer pays an annuity, referred to as the credit likelihood of a mortgage loan being prepaid in a given period, usually expressed viewed as tightening credit availability.
default swap spread, until the time of the credit event or the maturity date of on an annualized basis. Prepayments can occur when borrowers pay off their • Target set at FOMC (Federal Open Market Committee) meetings
the swap, whichever comes first. Expressed in annual percent of the FV.\ m o r t g a g e s e a r l y, r e fi n a n c e , o r s e l l t h e u n d e r l y i n g p r o p e r t y. Banking industry trends
- Off-balance sheet (SIVs) raise funds in the asset-backed commercial
It allows pure hedging, credit risk transfer to third parties and easy C P R = 1 − (1 − p )12, this is annualized
construction of a diversified portfolio of credit risks. paper market
Pass-through security: is the simplest mortgage-backed security: It represents - Invested in long-term assets
An add-up basket CDS pays off when any of the reference entities in the basket a claim to a fraction of the total cash flow that is flowing from the homeowners - Creates funding liquidity risk for banks via credit lines (liquidity
defaults. A first-to-default CDS pays off when the first default occurs. An nth-to- to the pool of mortgages. This simple structure implies that all investors in pass-
default CDS pays off when the nth default occurs. The value of these instruments backstop)
through securities are exposed to the prepayment risk. Note that the mortgage - I-banks increasingly financed themselves in the overnight repo market
depends on the correlation of defaults in the basket. pool pays WAC, but the holder of the pass-through security gets the pass-
SECURITIZATION (you borrow funds by selling collateral and agreeing to repurchase at
through rate rP T , which is lower 12 than WAC. The difference is the profit for
Origination: A bank or another financial institution creates or owns a portfolio later date)
whomever sets up the security. - I-banks needed to roll over a large part of their funding on a daily basis
of illiquid assets, such as mortgages 1 P (+ x b p s ) − P (− x b p s )
Pooling: These assets are grouped into a pool which can contain hundreds or Effective duration: D ≡ −
thousands of individual assets.
• P 2⋅xbps 2008 CRISIS
SPV (Special Purpose Vehicle): The pool of assets is transferred to an SPV, a P (+ x b p s ) and P (− x b s ) are the prices of the same security after we The 2008 financial crisis was triggered by the collapse of the subprime mortgage
separate entity created specifically to isolate the assets and associated risks from shift upward or downward the yield curve by x basis points, respectively and market and amplified by various mechanisms within the financial system. Here is
the originating institution's balance sheet. xbps is the shift. a detailed account of the key points in a more narrative form.
Issuance of Securities: The SPV issues financial securities backed by the assets 1 P (+ x b p s ) + P (− x b p s ) − 2 ⋅ P Premise: Why Structured Products
in the pool. These securities can take various forms, including bonds and asset- Effective convexity:C ≡ Structured products were used for several reasons. On the positive side, they
backed securities (ABS).
• P (x b p s )2 facilitated the transfer of credit risk to entities better equipped to manage it.
Sale of Securities: The securities issued by the SPV are sold to investors. The Negative convexity: If the interest rate decline results in an increase of the Banks held the riskiest tranches, incentivizing them to monitor loans closely.
cash flows generated by the underlying assets (such as interest and principal prepayment rate, then the price of the security does not increase by the same Pension funds invested in AAA-rated assets, while hedge funds focused on
payments from the mortgages) are used to pay the investors. amount as it would with a constant prepayment rate, because people are more riskier assets.
Collateralized Debt Obligations (CDOs): complex financial instruments that prone to pay refinance their debt, by paying the principal and in this way However, there were also less valid reasons for using these instruments. Many
aggregate and restructure various debts into a single tradable product. decreasing the cash flow of the MBS. banks engaged in regulatory arbitrage to circumvent Basel I regulations,
Advanced form of securitization widely used before 2008. Mechanically, the higher prepayment that is induced by lower manipulated credit ratings to achieve lower capital requirements, and relied
Tranching: The pool of assets is divided into tranches. Each one represents a interest rates pushes the pass-through price toward its outstanding principal heavily on the sometimes misleading assessments of rating agencies. This led to
different level of risk and return. The tranches are organized in a hierarchy based amount. overconfidence in structured products and risky investment strategies aimed at
on risk: Economically, the investor in a pass-through security has implicitly written enhancing portfolio returns through extreme tail risks and illiquidity
American call options to the homeowners of the mortgages in the pool.
• Senior Tranche: the safest and have payment priority. Investors in these exploitation.
tranches receive payments before others, but the returns are generally lower. (The graph, describing the variation in the price of security as interest rates Consequences
change, has an inflection point that ‘divides’ high interest rates, where we have a
• Mezzanine Tranche: Investors receive payments after the senior tranche The immediate consequences included an abundance of cheap credit and a
holders but before those in the subordinate tranche. positive convexity and consequently high price sensitivity to interest rates, and housing market boom. Lending standards deteriorated significantly, with the
low rates, where we have a negative convexity and consequently low price
• Equity Tranche: investors in these tranches receive payments only after all introduction of teaser rates, no-documentation mortgages, and NINJA loans (no
other tranches have been paid. In return, they can obtain higher returns. sensitivity to interest rates) income, no job, no assets), fueling a housing frenzy.
- CMOs offer different exposures to prepayment risk and the basic idea is to
Motivation of CDOs: Timeline of the Crisis
divide the total principal into smaller groups called tranche A, tranche B,
• possible more efficient allocation of credit risk (only 5% of bonds are AAA but Summer 2007: The increase in subprime mortgage defaults at the beginning of
many institutional investors are required to hold only highly rated bonds and tranche C and sequentially all the principal payments, scheduled or prepaid 2007 led to the closure of hedge funds and stress in the credit markets. Central
CDOs can be used to create AAA tranches even if none of the securities in the up to the point at which the whole payment of the tranche is paid out. For banks intervened by cutting rates and injecting liquidity.
underlying portfolio are AAA) example, Tranche A receives all principal payments until its principal is fully Fall and Winter 2007/2008: Banks began recording significant losses and
paid off, then Tranche B starts receiving principal payments, and so forth.
• It is used to remove loans from the balance sheet of banks, thereby reducing sought to raise new capital. The Federal Reserve (Fed) implemented a series of
the amount of required regulatory capital. In a synthetic CDO the actual Sometimes a “Tranche Z” is added, which receives no cash flows (like a zero rate cuts and created the Term Auction Facility (TAF) to address the liquidity
ownership of the loans is not transferred (default risk is transferred using coupon), but the coupon is accrued over time to its principal. Once all other crunch.
CDS). classes have been retired, the principal payments go to Tranche Z. Spring 2008: Credit spreads continued to widen, prompting significant
- Planned Amortization Class (PAC) securities are a type of tranche within a
• By retaining a significant fraction of the subordinated tranches, issuers can in interventions, such as the rescue of Bear Stearns by JPMorgan with Fed support.
principle mitigate moral hazard and adverse selection problems. CMO designed to provide predictable cash flows by offering protection Summer 2008: The situation worsened with the conservatorship of IndyMac
Correlation: Default correlation is a key input in the pricing of CDOs. Different against prepayment risk. PAC tranches receive payments according to a fixed and growing concerns about Fannie Mae and Freddie Mac, leading to
tranches have different exposures to default correlation. schedule based on prepayment assumptions, while the Companion tranche government interventions.
absorbs any deviations from these assumptions. Fall 2008: The collapse of Lehman Brothers marked a turning point, causing
• The value of the senior tranche depends negatively on default correlation - IO strips receive all interest payments from the underlying collateral and
• The value of the equity trance depends positively on default correlation severe market disruptions. Emergency measures were introduced, including
Issues related to the pricing of CDOs: none from the principal: If interest rates decline and prepayment increases, Henry Paulson's $700 billion bailout proposal, known as the Troubled Asset
lower interest payments will occur in the future as now less principal is
• Parameter sensitivity: the effects of errors in assumptions about the risks of Relief Program (TARP).
the underlying assets are amplified, particularly in CDO structures, correlation available to compute the interest, as a consequence the IO looses value. Winter and Spring 2008/2009: Various measures were implemented to stabilize
- PO strips receive all the principal payments, scheduled and unscheduled, and
parameter, default probability on individual bonds the economy, including economic stimulus programs and quantitative easing
no interest. If interest rates decline and prepayments increase, all cash flow
• Systematic risk: credit ratings are designed to give info about the expected operations to support the financial system.
payoff only. Hence they ignore the fact that CDO tranches give much higher will go to the PO strips, receiving cash flows in the immediate future, Amplification Mechanisms
exposure to systematic risk than single issuer corporate bonds of an equivalent increasing the value of the PO strips. The crisis was amplified by cascading effects on borrowers' balance sheets and
credit rating because they’re much more likely to default in bad times. Major players in the RMBS market loss and margin spirals. Investors, in order to maintain leverage, were forced to
Evidence suggests senior CDO tranches may not have offered high enough The three main players and their differences: sell assets, leading to further price declines. Additionally, lenders increased
- Ginnie Mae: Government National Mortgage Association (GNMA)
yields to compensate investors for this exposure to systematic risk. margin requirements, forcing borrowers to reduce leverage, exacerbating the
- Fannie Mae: Federal National Mortgage Association (FNMA)
BASICS OF THE MORTGAGE MARKET situation.
- Freddie Mac: Federal Home Loan Mortgage Corporation (FHLMC)
Mortgage: Homeowners borrow from banks to purchase their home. If the Credit Easing
borrower fails to remake the payment, the mortgage gives the lender the right HEDGING OF MBS PORTFOLIOS With interest rates near zero, the Fed expanded its toolkit by using the asset side
of foreclosure on the loan and he can therefore seize the property. Through Feedback from dynamic hedging of option positions on volatility in fixed income of its balance sheet to implement "credit easing" policies. This included lending
securitization the banks often sell these assets to other investors to raise capital. markets is a concern for policymakers. Mortgages have negative convexity: to financial institutions, providing liquidity backstops, and purchasing longer-
The market for mortgage backed securities serves the important role of Interest rates ↓, prepayment risk ↑ and buy treasuries, duration ↓ and interest term securities to support credit availability. !Until 2008, the Fed's balance sheet
transferring risk from those who have it to those who are better able to bear it, rates ↓. (assets) contained mainly traditional security holdings and amounted to 0.8
investors. The Federal Reserve’s decision to hold borrowing costs steady into trillion. From the crisis onwards, quantitative easing started by enlarging the
Distinction between a mortgage and a bond: 2015 and buy mortgage debt each month is reducing bond market volatility balance sheet by including Fed Agency Debt Mortgage-Backed Securities
• In a regular bond, the periodic coupon comprises only the interest of the and demand for options that hedge against changes in interest rates. Purchases, Liquidity to Key Credit Markets, Lending to Financial institutions and
bond’s principal, while the principal itself is repaid at maturity The combination of the strengthened forward rate Long Term Treasury Purchase in large quantities.
• In mortgage, the principal is repaid during the life of the mortgage together guidance from the Fed and a new round of quantitative easing Credit easing policy tools:
with the interest. Indeed, the coupon C contains 2 components: is a negative for volatility’’, said Ruslan Bikbov, a fixed-income Lending to financial institutions: – Discount window. – TAF to overcome
- interest payments strategist in New York at Bank of America Corp. ‘‘The perception of a longer Fed- stigma problem. – Shortage of Treasuries as collateral – TSLF and PDCF to
- principal repayment on-hold and the fact that the central bank is taking negative convexity out of the expand type of collateral.
• The fraction of the total coupon that is related to the interest payment and the mortgage-backed securities market will cause volatility to decline. Providing liquidity to key credit markets – Liquidity not transferred to credit
principal repayment varies over time -> Easier to hedge because less convexity markets. – Lending to financial institutions did not address strains in non-bank
• The reason is that the interest amount paid is determined by the amount of ROLE OF CENTRAL BANKS markets. – Liquidity backstop for money market funds. – CPFF—purchases 3-
the outstanding principal, which declines over time as principal payments Federal reserve act: month commercial paper to make sure financial institutions do not run into
occur. The larger the outstanding principal, the larger the amount of interest • maximum employment problems rolling over. – TALF—provide liquidity and capital to consumer and
that has to be paid. • stable prices: small business loan ABS market. – Maiden Lane I-III—tied to pools of assets that
It’s convenient to think about a mortgage as a special bond with only regular - precondition for maximum sustainable output growth, employment and Fed has lent against (Bear Stearns, AIG).
coupons, and no final lump sum principal payment. moderate long term interest rates. Purchasing longer-term securities. – Purchase of direct obligations of housing
Factors affecting refinancing: the general level of interest rates is an important - Prices are undistorted by inflation and serve as clearer signals and guides to related GSEs (Fannie Mae, Freddie Mac). – Purchase of MBS backed by Fannie
factor of prepayment. Other factors are: efficient allocation of resources. Mae and Freddie Mac—improve availability of credit for purchase of houses.
• Seasonality: summers are characterized by large prepayments, as this is the - Foster savings and investment. REGULATORY CHANGES
period for which people move from one place to another for various reasons • In the long run stable prices help foster economic activity The Dodd-Frank Wall Street Reform and Consumer Protection Act was signed
• Age of mortgage pools: young mortgages are characterized by large interest • Potential tension between the goals in the short run: into law in July 2010. The highlights of the Act are as follows: Identifying and
rate payments and low principal. By paying early, homeowners can save the - Example: upward pressure on prices as output is slowing down regulating systemic risk. Designate nonbank financial firms as systemically
interest rate payments. Because refinancing is costly, homeowners tend not to - Either control inflation but hurt output, or tackle unemployment but important, regulate them, and possibly break them up. Proposing an end to too-
refinance new or recently refinanced mortgages right away aggravate inflation big-to-fail. “Living wills” and orderly liquidation procedures for unwinding
• Family circumstances: default, disasters, or sale of the house - Policy makers face a dilemma systemically important institutions, ruling out taxpayer funding of wind downs. –
• Housing prices: if the property value of a house declines, it is more difficult to • Fed also contributes to financial stability by acting to contain financial Expanding the responsibility and authority of the Fed. Grants the Fed authority
refinance and thus prepayments tend to decline. disruptions over all systemic institutions and responsibility for preserving financial stability.
• Burnout effect: mortgage pools heavily refinanced in the past tend to be THE ROLE OF THE FED – Restricting discretionary regulatory interventions. Prevents or limits
insensitive to interest rates. Chances are that most or all of the homeowners Open market operations: emergency federal assistance to individual nonbank institutions. – “Volcker rule”.
that could take advantage or refinancing opportunities did so already in the • Purchases and sales of US treasury and federal agency securities. Limits bank holding companies to de minimis investments in proprietary trading
past. • Principal tool for implementing monetary policy. activities such as hedge funds and private equity, and prohibits them from
• The mortgage rate the bank receives is the return on capital on the • Determine the federal funds rate bailing out these investments. – Regulation and transparency of derivatives.
investment. The bank would like to receive this return for as long as possible. If • Objective is either quantity or interest rate Central clearing of standardized derivatives, regulation of complex derivatives
the mortgage owner has the choice of closing the mortgage, the bank will • OMO either permanent or temporary that can remain OTC.
miss this lucrative rate of interest. Prepayments bring down the rate of return Discount window:
for a bank: when the prepayment option is exercised by the mortgage owner, • Regional federal reserve bank’s lending facility for commercial banks and other
banks loose the stream of high interest payments that are now replaced by a depositary institutions
lower coupon on the same outstanding loan. The new mortgage coupon • Three discount windows programs are offered VaR + b
reflects the new, lower interest rate environment. Note that there is no default - Primary credit E S (99%) = [ (Φb − Φ( VaR )) + c (Φ(c ) − Φ(b ))] ⋅ 100
- Secondary credit 2
in prepayment, these are different concepts.
MBS - Seasonal credit
A number of similar mortgages are pooled together and serve as a collateral for • All discount window loans are fully secured
a MBS that is issued with face value equal to the cumulative outstanding • Primary credit was offered at premium of 100bps over the target rate, cut to
principal of the mortgages in the pool. MBS derive their characteristics from the 25bps above target in march 2008
features of the mortgage underlying pool. Three quantities are important in Reserve requirements:
determining the value of a mortgage backed security: • Reserve requirements are the amount of funds that a depositary institution
• Weighted Average Maturity (WAM): of the mortgages in the pool must hold in reserve against specified deposit
• Weighted Average Coupon (WAC): of the mortgages in the pool • Within limits specified by law, the Board of Governors has sole authority over
• Speed of prepayments: changes in reserve requirements
• Depositary institutions must hold reserves in the form of vault cash or deposits
with Federal Reserve Banks