4 Monte Carlo Methods

Monte Carlo simulations = simulation of random variables for the purpose of risk management - used for pricing complex financial instruments -> to build the distribution of portfolios that are too complex to model analytically - are quiet flexible and are becoming easier to implement with technological advancement in computing - Drawback: result depends heavily on models assumptions = shape of distribution, parameters, pricing function Simulations with one random variable - for creating artificial random variables with probabilities similar to risk factors of portfolio (e.g. stock prices, exchange rates, bond yields, bond prices, commodity prices) Simulating Markov Processes In efficient markets prices show a random walk pattern -> they follow a Markov process = a stochastic process independent of its past history = the entire distribution of future prices rely on the current price only, the past history is irrelevant Wiener process = describes a variable z, whose change is measured over the interval t such that its mean change is zero and variance proportional to t : z ~ N(0, t) z = t (if is the standard normal variable N(0,1))

increments z are independent over time Generalized Wiener process = describes a variable x built up from a wiener process, with in addition a constant trend a per unit of time and volatility b: x= a t+ b z particular case: martingal = a zero drift stochastic process (a = 0) which leads to E(x) = 0 -> expectation of a future value = the current value The Ito process = describes a generalized Wiener Process, whose trend and volatility depend on the current value of the underlying variable and time: x= a(x,t)t+ b(x,t) z -> a Markov process because the distribution depends only on the current value of the random variable x, as well as time -> in addition the innovation in this process has a normal distribution. The Geometric Brownian Motion (GBM) = particular example of an Ito process which describes the variable S as S = St+ S z
expected total rate of return on the asset minus the rate of income payment, or dividend yield (stocks)

Seite 1 von 64

-> process is geometric because trend and volatility term are proportional to the current value of S -> typical case for stock prices (rate of return appears more stationary than raw dollar return) -> also used for currencies because S/S represents the capital appreciation only, abstracting from dividend payments model is important because it is the underlying process for the Black Scholes formula. -> key feature = the fact that the volatility is proportional to S = insurance that stock price stays positive (stock price falls -> variance decreases -> unlikely to experience a large down move which would push price into negative values) limit of this model = a normal distribution for dS/S = dln(S) -> S follows a lognormal distribution implies that over an interval T-t = the logarithm of the ending price is distributed as ln(ST) = ln(St) + ( / 2) +
a standardized normal variable.

whether normal or lognormal distribution is seen as better depends on horizon considered - one day only (no real difference) very unlikely to drop below zero given typical volatilities -> choice does not matter - measurement in years lognormal much more realistic (no negative prices) simulation approximates this process with small steps with a normal distribution with mean und variance given by S/S ~ N(t, t ) -> to simulate future price path for S: start with current price St and generate a sequence of independent standard normal variables for i = 1,2,.n. -> next price S t+1 is built asc+St (t+ t x 1) and so on until target horizon is reached. there S t+n = ST should have a distribution near to lognormal experiment can be repeated as often as needed K = number of replications shortcomings of model: Price increments are assumed to have a normal distribution (in real world fat tails !!) Returns may experience changing variances Jumps are not simulate able Drawing Random Variables most spreadsheets or statistical packages provide functions that can create uniform or standard normal random variables -> can be easily extended to distributions that better reflect the data (e.g. fatter tails, non-zero skewness) -> methodology involves inverse cumulative probability distribution function Example normal distribution: cumulative probability distribution function N(x) always is between 0 and 1 -> analytical formula can be inverted easily 1. Generate a uniform random variable u drawn from U(0,1). 2. Compute x that u = N(x) or x = N -1(u) e.g. u =0,0430 x = -1,717 because u < 0,5 -> x<0 Variable can be transformed into any normal variable by multiplying by the standard deviation and adding mean. a distribution function can be generated as long as the cumulative distribution function is inverted

Seite 2 von 64

Simulating yields GBM used for stock prices and currencies -> fixed-income products are another matter bond prices display a long-term reversion to the face value, which represents the repayment of principal at maturity (assuming no default) -> process is inconsistent with GBM, which displays no such mean reversion -> volatility of bond prices also changes in a predictable fashion as duration shrinks -> commodities also display mean reversion taken into account by modeling bond yields directly in a first step -> in a next step, bond prices are constructed from the value of yields and a pricing function -> dynamics of interest rates rt can be modeled by (one-factor model: only one stochastic variable): rt = K(- rt) t+ rtY zt
zt ..Wiener Process 0 < K < 1, >= 0, >= 0

Features: - Mean reversion to a long run value of : K governs the speed of mean reversion -> when interest rate high > model creates a negative drift, low current rates create a positive drift - volatility process: model includes Vasicek model (= changes in yields are normally distributed because r is a linear function of z) when = 0 -> leads to closed-form solutions for many fixed-income products -> but might lead to negative interest rates when initial rate starts with a low value (volatility of the change rate does not depend on the level unlike in the GBM) -> with Y = 1 this is the lognormal model -> with Y = 0,5 this is the Cox, Ingersoll and Ross (CIR) model this class of models is known as equilibrium models: start with an assumption about economic variables and imply a process for short term interest rate r -> generate a predicted term structure, whose shape depends on model parameters and initial short rate are not flexible enough to provide a good fit for todays term structure second type of models: no-arbitrage models: term structure is input for parameter estimation -> earliest model: Ho and Lee model rt = (t)t+ zt
(t) function of time chosen so that model fits the initial term structure

-> was extended to incorporate mean reversion in the Hull and White model rt = [(t)-a rt ] t+ zt -> Heath, Jarrow and Morton model: go one step further assume that volatility is a function of time Cons of no-arbitrage models: do not impose any consistency between parameters estimated over different dates -> (t) could be different from day to day which is illogical -> more sensitive to outliers or data errors in bond prices Binominal trees Simulations are useful to mimic uncertainty in risk factors (esp. with numerous) -> sometimes also useful to describe the uncertainty in prices with discrete trees

Seite 3 von 64

 tree is called binominal when the price can take one of two steps -> binominal model = discrete equivalent to GBM -> horizon T is divided in n intervals -> at each node the price is assumed to go up by a probability of p or down by the probability of (1-p) -> u,d,p are chosen that for a small time interval the expected return and variance equal those of the continuous process U = e t, d = (1/u), p = e t d / (u-d) This matches the mean e t d E[S1 / S0 ] = pu + (1-p)d = u-d *u+ u-d u- e t *d = u-d e t(u-d)-du+ud = e t

as the number of steps increase the discrete distribution converges to lognormal distribution Implementing Simulations Simulation of VAR implementing of MC methods for risk management follows the steps - Choose stochastic Process for risk factor price S (e.g. distribution) - Generate pseudo-random variables representing risk factor at target horizon - Calculate value of portfolio at the horizon - Repeat step 2 and 3 as often as necessary These steps create a distribution of values , which can be sorted to derive the VAR measure the quantile and the average value VAR = average deviation from the expected value on the target date
Ave(FT) average value Q(FT,,c) quantile

Simulation for Derivatives focus on expected value on the target date discounted into the present -> valuation focuses on the discounted center of the distribution (cf. VAR focuses on the quantile on target date) MC long used for pricing derivatives -> pricing can be done by assuming that underlying grows at risk-free rate r (assuming no income payment) e.g. pricing an option on a stock with expected return of 20% is done in assuming that stock grows at risk-free rate of 10% and by discounting the same risk-free rate = risk-neutral approach. contrary risk management deals with actual/physical distributions -> for measuring VAR, risk manager must simulate asset growth by using actual expected return risk management uses physical distributions, pricing uses risk-neutral distributions these methods are not applicable to all type of options -> assume that value of instrument can be priced as solely function of end-of-period price and perhaps of its sample path (e.g. Asian option = payoff is function of the price averaged over sample path option is pathdependent) simulations are inadequate to price American options (early exercise) -> optimal exercise decision too complex for model (future values) not with regular simulation methods which Seite 4 von 64

consider only present and past information. -> they need backward recursion (e.g. binominal trees) -> examines whether option should be exercised or not, starting from the end working backward until the starting time Accuracy effect of sampling variability: unless K is extremely large the empirical distribution of ST will only be an approximation of the true distribution -> natural variation in statistics measured from MC. MC involves independent draws -> standard error is related to square root of number of draws much more simulations will increase precision but only at a small rate -> greater problem for risk management, because quintiles are estimated less precisely than the average. VAR: precision is also a function of selected confidence interval -> higher confidence = fewer observation in left tail = less precise VAR measures. -> shape of distribution also influences precision of quantile Relative to a symmetric distribution a short option has negative skewness (=long left tail) > observations much more dispersed difficult for VAR estimation various methods to speed up convergence: Antithetic variable technique = uses twice the same sequence of random draws -> takes original sequence and changes value (value*-1) twice the number of points in final distribution without running twice the number of simulations Control variable technique = used to price options with trees when a similar option has an analytical solution Quasi random sequences/ Quasi Monte Carlo (QMC) = creates draws which are not independent, but are designed to fill sample space more uniformly -> converge faster as MC (closer to the true value) -> traditional MC also provides a standard error, therefore idea of how far the estimated value might be from true value important for decision on number of replications Multiple Sources of risk more general case of simulation with many sources of risk -> if factors are independent the randomization can be performed independently for each variable (GBM model) -> in general risk factors are correlated, therefore the drawn independent variables need to be transformed into correlated ones The Cholesky Factorization generate N joint values that display the correlation structure of R -> because the matrix R is symmetric it can be decomposed into is so-called Cholesky factors. R=TT
T = is a lower triangular matrix with zeros on the upper right corners (above diagonal)

in practice this decomposition yields to a number of insights: it will fail if number of independent factors implied in the correlation matrix is less than N Curse of dimensionality modern risk management is about measuring risk of large portfolios, typically exposed to large number of risk factors. number of computations increase geometrically with N Dimension of covariance matrix = N(n+1)/2 Simplification of risk factors (use only significant ones) -> one way to do this is principalcomponent analyses (PCA) = finds linear combinations of risk factors that have maximal explanatory power Seite 5 von 64

5. Introduction to Derivatives
Derivate = financial contract traded private (over-the-counter OTC) or on organized exchanges -> derive their value from some underlying index (typically price of asset) classified in two categories: linear (forward, futures, swaps) instruments: value = linear function of underlying index -> are obligations to exchange payments according to specified schedule -> Forward(very simple to evaluate and price), as are futures (traded on exchanges), swaps much more complex nonlinear instruments (options): options traded both OTC and on exchange , value = nonlinear function of underlying Overview of derivative markets derivative instrument = private contract whose value derives from some underlying asset price, reference rate or index such as a stock, bond, currency or commodity contract must specify a notional amount (=principal), which is described in terms of currency, shares bushels, or some other unit -> movements in value depend on notional and underlying contracts/ private agreements between two parties -> sum of gain and losses = 0 (one side gains, other looses same amount) -> cf. securities (bonds, stocks): are issued to raise capital Classification of derivatives markets by underlying and type of trading (=broadest level) interest rate contracts = most wide spread type -> on OTC also currency contracts (in forwards and forex swaps = combination of spot and short time forward transaction) -> exchange show interest futures and options as most common notional amount several times the world GDP (gross domestic product) and also higher tan total outstanding value of stocks Forward contracts most common transactions in financial instruments are spot transactions -> physical delivery as soon as practical -> beneficial to trade for delivery at some future date -> hedge out price fluctuation forwards = private agreement to exchange a given asset against cash or other asset at a fixed point in the future -> terms of contract: quantity, date and price long = position which implies buying, short = position to sell represent contractual obligations: exchange must occur whatever happens to the intervening price (unless default) -> no choice in taking delivery or not hedge against movement in price = locking price for delivery now face/notional amount = principal value = amount to pay at maturity assumption of continuously compounded interest rates Value of forward contract at expiration = VT = ST-F (for one unit of the underlying asset) Value of contract at expiration is derived from the purchase and physical delivery payment of cash in exchange for actual asset Second mode of settlement = cash settlement = measuring market value of asset upon maturity and agreeing for the long side to receive nV T = n(ST-F) -> amount is positive =profit or negative = loss Seite 6 von 64

Payoff of profits on Long Forward Contract payoff

Payoff of profits on Short Forward Contract payoff

Sport price of underlying asset

Sport price of underlying asset

Valuing Forward Contracts two questions when evaluating forward contracts 1) How is current forward price determined? 2) What is current value of an outstanding? initially assumption = forward pays no income, no transaction costs = no bid-ask spread, and ability to lend and borrow at same risk-free rate. generally forward contracts are established that initial value = 0 through setting forward price by a no-arbitrage relationship between cash and forward markets Arbitrage = zero-risk, zero-net investment strategy Strategies I) Buy one share at spot price St and hold it to time T II) Enter forward contract to buy, sell one unit of same underlying asset at forward price.Ft -> in order for sufficient funds at maturity = investment in an interestbearing account Ft = Stert -> forward rate higher than spot rate (reflects that there is no down payment to enter the forward contract and also reflects the time value of money) -> mispriced forwards lead to arbitrage (buy cheap asset and sell the expensive one) !!!! Valuing on Off-Market Forward contract Use arbitrage reasoning to evaluate outstanding forward contract -> off-market: K differs from prevailing forward rate Market value of outstanding long position Vt = St-Ke-rt gains value when underlying increases in value Valuing on Off-Market Forward contract Assets producing no income payment assumed -> but in real: Stocks that pay regular dividend Bond that pays regular dividend Stock index that pays dividend stream approximated by continuous yield Foreign currency that pays foreign-currency denominated interest rate Income: discrete = fixed amounts at regular points in time, continuous = accrued in proportion of time asset is held -> assumption: fixed and certain payments prices are measured in the domestic currency Fte-rt =St-PV(D)
PV(D) present value of dividends/coupon payment -> discounted at the risk-free rate St stock price at t

Seite 7 von 64

Fte-rt =St e-r*t

for continuous payment -> known as interest rate parity when dealing with currencies

r* < r -> Ft > St: asset trades at a forward premium r* > r -> Ft < St: asset trades at a forward discount forward price differs from the spot price to reflect the time value of money and the income yield on the underlying asset -> is higher than the spot price if the yield on the asset is lower than the domestic risk-free interest rate, and vice versa. current value of an outstanding forward contract can be found by entering an offsetting forward position and discounting the net cash flow at expiration Value of outstanding forward with income payment Vt = Ste-rt - Ke-rt Vt = (Ft K)e-rt

Futures forward contracts allow users to take positions that are economically equivalent to those in the underlying cash market, but do not involve substantial up-front payments forwards have leverage futures contracts are standardized, negotiable, and exchange-traded contracts to buy or sell underlying asset Differences from forwards: Trading on organized exchanges Standardization: limited choice of expiration dates, fix contract sizes large secondary market = easy to buy and sell (good liquidity) -> but also less precisely suited to need of some hedgers (creates basis risk) Clearinghouse = standardization of counterparty (ensures the performance of the contract, bears credit risk) Marking-to-market= settlement of gains and losses daily -> no large losses over time Margins = upfront posting of collateral that can be seized should the other party default -> if equity below maintenance margin, customer must provide additional funds to cover the initial margin -> level depends on the instrument and type of position Valuating Future contracts is similar to forward contracts, main difference is that any profit or loss accurse during life of futures against all at once at expiration -> correlation = 0: no difference whether payments are received earlier or later -> futures price = forward price -> interest-rate futures: value is negatively correlated to interest rates -> long futures price must be lower than the forward price (convexity effect) Swap contracts =OTC contracts to exchange a series of cash flows according to prespecified terms -> usually longer periods as futures or forwards, principal = notional principal if amount is in same currency -> no need to exchange principal can be viewed as a portfolio of forward contracts -> priced using valuation formulas for forwards

Seite 8 von 64

6. Options
options = non-linear derivatives Plain vanilla options (puts, calls) Exotic options Option payoffs Basic options Option = instrument that gives holder the right to buy or sell an asset at a specified price until a specified expiration date -> Delivery price = exercise price = strike price Options to buy = Call options, Options to sell = Put options will only be executed if generate profit -> sellers are said to write options Depending on time of exercise: - European option: only at maturity - American Option: any time payoff profile of long calls = CT = Max(ST-K,0), payoff of long put = PT = Max(K-ST, 0)
K exercise price ST asset price at T

Current price close to strike price = at the money Current price so that option will be exercised = in the money = intrinsic value large - call: St > K - put: St < K Current price so that option will not be exercised = out of the money Payoff diagramms at exercise time (hockey stick diagrams)

Long Call

Long Put

Short Call

Short Put

buying options can generate only profits (or 0) at expiration -> payment is needed to acquire the contract (as it always has value >= 0) -> up-front payment = option premium option becomes more expensive as it moves in-the-money for a given spot price the sum of the profit and loss for the long and for the short = 0 long position has limited downside risk = loss of premium short position has unlimited downside risk = no upper limit on S (worse loss: S = 0) options can also be struck on futures -> exercising a call = investor becomes long the future contract, exercise a put = creates short position in the future contract

Seite 9 von 64

Put-Call Parity Option payoffs can be used to build more complex positions long Position in asset = long call (provides the equivalent of the upside risk) + short put (generates the same downside risk as holding the asset) link between value of call and value of put = put-call-parity only European options are considered with the same maturity and exercise price Portfolio consists of long call and short put and investment to be sure to be able to pay exercise price at maturity c-p= Se-r*T-Ke-rT = (F-K)e-rt A long position in an asset is equivalent to a long position in an European call with a short position in an otherwise identical put, combined with a risk-free position.

Combinating Options Multiply combinations possible, with each other or with underlying asset covered call = long position in the stock & short call (to collect the premium)

protective put = long position in the stock & short put (to protect the downside) collar = if long in the stock: buy long put (low strike) & short call (high strike) -> equivalent to a short stock position -> create net payment of 0 straddle = call & put (strike price and maturity same) -> long straddle benefits from large price moves whether up or down

strangle = call & put (different strike prices) -> are out-of-the-money and therefore cheaper than straddles spreads = positions with only one type of options o Calendar/horizontal spreads =different maturities o Vertical spreads =different strikes o Diagonal spreads = move across maturities and strikes o Bull spread = advantage in increase of underlying asset price

Seite 10 von 64

o Bear spread = represents bet on a falling price o Butterfly spread = three types of options with same maturity -> 1 long call, two short calls and another long call ( K1<K2<K3 ) -> position benefits when the underlying asset price is stable (close to K2)

o Sandwich = opposite to a butterfly spread Option premiums instantaneous relationship between option value and current price S: - call: a higher price S increases the value of the option (non-linear, convex) - put: a lower price S increases the value of the option (non-linear, convex) hockey-stick only at time of maturity

premium decomposed in: - Intrinsic value = value of option if exercised today -> call: MAX(St-K, 0), put: MAX(K- St, 0) - Time value = consist of remainder which reflects the possibility of future gains -> increases with the volatility of the underlying asset options can be classified: - at-the-money: current spot price is close to the strike price - in-the-money: intrinsic value is large - out-of-the-money: spot price is much below the strike price for calls, conversely for puts -> intrinsic value = 0 European option: general bounds (otherwise arbitrage) - current Value of call <= asset price - value of a European call >= price of asset PV of the strike price Seite 11 von 64

- current value of a put <= K - lower pound of the put: value >= PV of strike price price of asset American options: early exercise basic trade-off arises between value of the American option dead = exercised and alive = not exercised -> choice between exercising or selling on the open market American call option on a non-dividend-paying stock (or asset with no income) should never be exercised early -> if the asset pays income, early exercise may occur, with a probability that increases with the size of income payment -> only exercise call early to capture a dividend payment -> American call: exercise early = St-K -> is > St-Ke-rt: value of the option alive must be worth more than the equivalent European call a high income payment makes holding the asset more attractive than the option -> American options on income-payment assets may be exercised early -> also applies to options on futures (implied income stream on the underlying = risk-free rate) American puts: Pt >= K - St -> is > Ke-rt - St (= lower bound for European puts) -> American puts on non-income paying assets could be exercised early -> less attractive to sell the asset when lower interest rates and with higher income payments on the asset Valuing Options Pricing by Replication: Philosophy of pricing models = replicate payoff on the instrument (option) by a suitable portfolio of the underlying asset plus a position (long or short) in a risk-free bill -> no arbitrage: price = price of replication portfolio First possibility by binominal process: -> Initial price = S can only move to two different end prices (up or down) current value of the stock S0 = [p x S1 + (1-p) x S2]/(1+r) discounted expected payoff assuming investors are risk-neutral -> solve to calculate the risk-neutral probability p value of the option C0= [p x c1 + (1-p)x c2]/(1+r) concept of risk neutral pricing

Black Scholes Valuation (BS) Provides elegant closed form solution to the pricing of European calls based on 4 assumptions: the price of underlying asset moves in a continuous fashion interest rates are known an constant variance of underlying asset returns is constant capital markets are perfect (short-sales, no transaction costs, markets operate continuously ) statistical process of asset price is modeled by a geometric Brownian motion (over short time the logarithmic return has a normal distribution) total return dS/S = dt + dz
dt represents the drift component dz stochastic component

-> dz normally distributed with mean = 0 and variance dt -> price is lognormal distributed Seite 12 von 64

 position in the option can be replicated by a delta position in the underlying asset -> portfolio combining asset and option in appropriate proportions is locally risk-free risk neutrality: in risk-free world expected return on all securities = risk-free rate of interest, no risk premium for inducing to take risks -> artificial method to obtain correct solution no implication that investors are risk-neutral! limited use for risk management -> BS can be used to derive the risk-neutral probability of exercising the option -> For risk management physical probability (= actual probability of exercise) much more important and differs from risk-neutral probability Extensions Merton expanded BS to the case of stock paying continuous dividend Garmann-Kohlhagen-model: extends formula to foreign currencies reinterpreting yield as foreign rate of interest -> call that is deep in-the-money is equivalent to a long forward contract (almost certain exercised) Black model: applies formula to options on futures -> with an option on cash, the income is the dividend or interest on the cash instrument -> with an option on futures, the implicit income is the risk-free rate of interest -> future can be viewed as equivalent to a position in the underlying asset with the investor setting aside an amount of cash equivalent to the present value of the future Standard options involve a choice to exchange cash for asset (special case of exchange option = surrender an asset in exchange for acquiring another) Margrabe has shown that valuation formula is similar to the usual model Market versus Model Prices in practice BS is used to price options -> all parameters are observable, except volatility observing market prices: we can solve for volatility parameter that sets the model price equal to the market price = implied standard deviation (ISD) -> Model = correct ISD should be constant for different strikes -> but this is not the observation: plots of ISD against strike display volatility smile pattern (= ISDs increase for low and high values of K) - Negative slope is called volatility skew related concept is term structure of volatility: refers to observation that ISD differs across maturities -> arises because option market incorporates many types of events over the lifetime of the option (e.g. realized volatility of a stock tends to increase on the day of an earnings announcement) observations summarized by an implied volatility surface = three dimensional plot of ISD across maturities and strike prices -> traders use heuristic approaches to extrapolation of curve over investment horizon First scenario = sticky strike = curve does not change and ISD drops -> assumes no structural change in volatility curve, price movement largely temporary Second scenario = sticky moneyness = curve shifts to the right and ISD stays the same -> assumes permanent shift in the volatility curve ISD of a portfolio of assets can be related to the ISD of its components through the implied correlation Usually portfolio variance is related to the individual volatilities

Seite 13 von 64

-> increasing correlation = increasing total portfolio risk Other Option Contracts Binary/digial options = pay fixed amount if asset price ends above the strike price -> involves a sharp discontinuity around strike (just below K: value = 0, just above: value = notional) difficult to hedge

Barrier options: payoff depends on value of asset hitting barrier during a certain period of time Knock-out option: disappears if the price hits a certain barrier (down-and-out call: disappears when S = H < S0, up-and-out call: disappears when S = H > S0) Knock-in option: comes into existence when price hits barrier (down-and-in call: appears when S = H > S0, up-and-in call: appears when S = H < S0) -> down-and-out and down-and-in are perfectly complementary with identical parameters (add up to a regular call) -> sometimes the options offer a rebate if its knocked out -> similar combinations for put (up and out put: disappears when S = H > S0, down and out put: disappears when S = H < S0) -> option is exercised at maturity if S < K -> are attractive because cheaper than equivalent European option -> less likely to be exercised -> but difficult to hedge Asian Options/ Average rate options: generate payoffs that depend on the average value of the underlying spot price during life instead of the ending value -> cheaper (lower volatility), easy to hedge Chooser options: allow holder to choose whether option is call or put -> package of two options: regular call + option to convert to a put -> more expensive than plain-vanilla options Look back options: payoffs that depend on the extreme values of S over option lifetime -> much more expensive Valuing options by numerical methods Some options have analytical solutions (e.g. BS for European vanilla options) -> for more general options numerical methods have to be used Basic valuation states that current value is discounted present value of expected cash flows, where asset grow at risk-free rate and discounted with same risk-free rate -> use MC to generate sample paths, final option values and discount them to present such simulation methods can be used for European or even path-dependent options (e.g. Asian options) Simulation cannot account for possibility of early exercise -> binominal trees for American options

Seite 14 von 64

7 Fixed-Income Securities
Fixed-income securities = bonds that promise to make fixed coupon payments (primary definition) -> evolved to include any securities that obligates the borrower to make specific payments to the bondholder on specific dates Bond = security that is issued in connection with a borrowing arrangement -> in exchange for received cash, the borrower becomes obligated to make a series of payments to the bondholder Fixed-income derivates = instruments whose value derives from some bond price, interest rate or other bond market variables Securitization = the process by which assets are pooled and securities representing interest in the pool are issued Overview of debt markets fixed income markets are truly global broad classification by borrower and currency type domestic bonds = bonds issued by resident entities and denominated in the domestic currency foreign bonds = bonds floated by a foreign issuer in the domestic currency and subject to domestic country regulations Eurobonds = mainly placed outside the country of the currency in they are denominated and are sold by an international syndicate of financial institutions Foreign bonds and Eurobonds constitute to the international bond market Global bonds = bonds that are placed at the same time in the Eurobond and one or more domestic markets with securities fungible between these markets domestic bond market: government bonds/ sovereign bonds = issued by the governments Brady bonds = sovereign bonds issued in exchange for bank loans -> hybrid securities whose principal is collateralized by U.S. Treasury zero-coupon bonds (no risk of default on the principal) Tesebonos = dollar-denominated bills issued by the Mexican government government agency and guaranteed bonds = issued by agencies or guaranteed by the central government which are public financial institutions state and local bonds /mutual bonds = issued by local governments bonds issued by private financial institutions (incl. Banks, insurance companies, issuers of asset-backed securities) corporate bonds = issued by private nonfinancial corporations (incl. Industrials, utilities) debt securities market bond markets = fixed-income securities with remaining maturities beyond one year money markets = fixed-income securities with remaining maturities below one year

Seite 15 von 64

 Mortgage-backed securities (MBS) = securities issued in conjunction with mortgage loans (= loans secured by the collateral of a specific real estate property) -> payments on MBSs = repackaged cash flows supported by mortgage payments made by property owners can be issued by government agencies or private financial corporations Asset-backed securities (ABS) = securities whose cash flows are supported by assets such as credit card receivables or car loan payments Fixed-income Securities interest payments on a regular basis: U..S. Treasury and corporate bonds = semiannual, Eurobonds = annual, some others = quarterly Instrument types Fixed-coupon bonds = pay a fixed percentage of the principal every period and the principal as a one-time payment at maturity (balloon) Zero-coupon bonds = pay no coupons but only the principal, return is derived from price appreciation only Annuities = pay a constant amount over time which includes interest plus amortization of the principal Perpetual bonds and consols = no set redemption date, value derives from interest payment only Floating-coupon bonds/ floating-rate notes (FRNs) = pay interest equal to a reference rate plus a margin, reset on a regular basis Structured notes = more complex coupon patterns inverse floaters/reverse floaters = have coupon payments that vary inversely with the level of interest rates (best in falling interest rate environment) Inflation-protected notes = principal is indexed to the Consumer Price Index (CPI), provide protection against an increasing rate of inflation additional variations: step-up bonds = have fixed coupons that start at a low rate and increase over time benchmark interest rates LIBOR = London Interbank Offer Rate = reference rate based on interest rates at which banks borrow unsecured funds from each other in the London interbank market -> published daily by the British Bankers Association around 11:45 a.m.LT, computed from an average of the distribution rates provided by reporting banks -> calculated from 10 different currencies and various maturities (overnight to one year) EURIBOR = Euro Interbank Offer Rate = sponsored by the European Banking Federation, published by Reuters at 11:00 a.m CET. EONIA = Euro Overnigh Index Average = onvernight unsecured lending rate (by the some panel of banks as the EURIBOR) -> published every day before 7:00 p.m. Bonds with option features: callable bonds = issuer has the right to call back the bond at fixed prices on fixed dates -> call back the bond when the cost of issuing new debt is lower than the current coupon paid on the bond pricing of the option feature: decompose into a long position in a straight bond minus a call option on the bond price -> unfavourable for investors who require a lower price to purchase the bond, thereby increasing its yield puttable bonds = investor has the right to put the bond back to the issuer at fixed prices on fixed dates -> dispose of the bond should its price deteriorate -> put option makes the bond more attractive, increasing its price and lowering its yield Seite 16 von 64

convertible bonds = bon can be converted into the common stock of the issuing company at a fixed price on a fixed date -> partake in the good fortunes of the company, allows companies to issue bonds at a lower yield than otherwise

Methods of Quotation Clean price: without accounting for the accrued income from the last coupon Gross price/ dirty price: clean price + accrued interest -> computed as act/act in the U.S. Treasury market, act/360 in the LIBOR market actual number of days since last coupon Accrued Interest (AI) = Coupon * Actual number of days between last and next coupon Discount basis for Treasury bills: quoted in terms of an annualized discount rate Discount rate (DR) = (Face Price)/ Face * 360/actual number of days Price (P) = Face * (1 DR * t /360) Face/Price = (1 + yield * t/365) If the price is lower than the face value, the yield must be greater than the discount rate Analysis of fixed-income Securities The Net Present Value (NPV) Approach
Ct cash flow in period t (coupon and/or principal payment t number of periods to each payment T number of periods to final maturity y yield to maturity P price of the bond, including accrued interest

yield = another way to express the price of the bond, more convenient to compare various bonds -> is the expected rate of return on the bond. Provide all coupons are reinvested at the same rate Pricing fair value of the bond: discount the cash flows using the market-determined yield which does not depend on t but is fixed for all payments for the bond -> ignores the shape of the term structure of interest rates
YT market determined yield

short maturities have lower interest rates than longer maturities -> discount each cash flow at the zero-coupon rate (sport interest rate) that corresponds to each time period -> add a fixed amount/ static spread to the spot rate so that the NPV equals the current price (also possible as yield spread but less accurate) bond with a large static spread is cheaper or has a higher expected rate of return
Rt spot interest rate for maturity t and this risk class (same currency and credit risk SS static spread

Seite 17 von 64

 cash flows with different credit risks need to be discounted with different rates -> separate the discounting of principal from the discounting of cash flows principal is discounted into a present value using the appropriate yield present value is subtracted from the market value coupons are discounted at the stripped yield which accounts for the credit risk of the issuer Duration = measure of exposure/sensitivity of the bond price to movements in yields = the weighted average time to wait for each payment, but only when the bonds cash flows are predetermined fixed cash flows: duration is measured as the weighted maturity of each payment, where the weights are proportional to the present value of the cash flow Macaulay duration

Dollar duration (D* P) = expressed in dollar value of a basis point (DVBP)

D* modified duration

floating-rate notes: - just before the reset date the prevailing interest for setting the coupon is known -> FRN is similar to cash/money market instrument with no interest rate risk and selling at par with zero duration - just after the reset date the investor is locked into a fixed coupon over the accrual period -> FRN is equal to a zero-coupon bond with maturity equal to the time to the next reset date Spot- and Forward Rates Spot rates = zero-coupon investment rates that start at the current time Forward rates = are rates that start at a future date -> can be inferred from spot rates = the break-even future rate that equalizes the return of investment of different maturities -> can be interpreted as a measure of the slope of the term structure and can be viewed as an expectation of the future spot rate (expectation hypothesis: assumes that there is no risk premium embedded in forward rates and that it does not matter which maturity should be selected for investment purposes -> exchange: higher coupons <-> capital loss)

Par yield = can be viewed as a weighted average of spot rates

Seite 18 von 64

 when the spot rate curve is flat, the spot rate curve is identical to the par yield curve and the forward rate curve as interest rates have to be positive, forward rates have to be positive -> otherwise there would be arbitrage Prepayment Factors affecting mortgage refinancing patterns spread between the mortgage rate and the current rates: increase in spread increases prepayments age of the loan: low prepayment rates after the issuance of the loan seasoning: increasing of prepayment level until it reaches as stable level refinancing incentives: smaller costs of refinancing makes home-owners refinance more often previous path of interest rates: refinancing is more likely when rates have been high in the past but recently dropped burnout: if rates are low and have been for a while most if the principal is already prepaid level of mortgage rates: lower rates increase affordability and turnover economic activities: greater job turnover leads to prepayments seasonal effects: more home-buying in spring, more prepayments in fall
SMM single monthly mortality CPR conditional prepayment rate

Public Securities Association prepayment model: industry standard to describe the schedule of projected prepayments during the remaining life if the bond Prepayment risk mortgages are subject to market risk (e.g. fluctuations in interest risk) and credit risk (e.g. home-owner default) prepayment risk = risk that the principal will be repaid early - contraction risk: = average life of the loan is shortened (decreasing interest rates) - extension risk = average life of the loan is lengthened (increasing interest rates) -> mortgage investments have negative convexity which reflects the short position in an option granted to the home-owner to repay early option feature increases the yield option adjusted spread = most commonly used approach to model the option component -> starting from the static spread -> running simulations of various interest rate scenarios and prepayments to establish the option cost -> with the same risk class, a security trading at a

Seite 19 von 64

high OAS is preferable to others -> more stable over time than the spread (which is affected by the option component) Option Adjusted Spread (OAS) = Static spread Option cost effective duration/effective convexity -> measures are based on the estimated price of the mortgage for three yield values, making suitable assumptions about how changes in rates should affect prepayments

Securitization Principles of securitization mortgage loans are not tradable Securitization = the process by which assets are pooled and securities representing interests in the pool are issued Mortgage-backed Securities (MBS) = tradable securities that represent claims on pools of mortgage loans -> often with third party guarantees against credit risk (Fannie Mae: government sponsored enterprise (GSE) as mortgage insurer) Process of securitization: 1) create a new legal entity = Special-purpose vehicle (SPV)/ Special-purpose entity (SPE) 2) originator/issuer pools a group of assets and sells them to the SPV 3) SPV issues tradable claims or securities that are backed by the financial assets if clean sale to the SPV: shields the ABS investor from the credit risk of the originator pooling offers ready-made diversification across many assets banks as financial intermediaries = raise funds (recorded as liabilities on the balance sheet) that are used for making loans (recorded as assets) securitization removes both assets and liabilities from the balance sheet requiring less equity capital, regulatory capital relief for the originator its an additional source of funding manage the banks risk profile (lowering the risk profile with securitization if collateral is much riskier than the rest of the assets) Assets included in ABS: collaterals: mortgage loans, auto loans, student loans, credit card receivables, accounts receivables, debt obligations Servicing agent = collecting payments on the collaterals -> usually also performed by the originator in exchange for a servicing fee Seite 20 von 64

Pass-through = one class of bonds and all investor receive the same proportional interests in the cash flow -> tranches = bonds, when the SPV issues different classes of securities participation certificates:= mortgage pass-through with GSE as mortgage insurer On-balance sheet securitization = covered bonds/Pfandbriefe the bank originates the loans and issues securities secured by these loans which are kept on its books -> investors have recourse against the bank in the case of default on the mortgages, bank provides a guarantee against credit risk Residential mortgage-backed securities (RMBS) = collateral consists of residential mortgage loans Commercial mortgage-backed securities (CMBS) = collateral consists of commercial mortgage loans cash flow pattern starts from an annuity, where the home-owner makes a monthly fixed payment that covers principal and interest subject to interest rate risk, prepayment risk and default risk Issues with securitization less incentives to worry about the quality of the loans because its revenues depend on the volume of issuance (moral hazard -> institutions behave differently than if fully exposed to the risk of the activity) problem when a bank decides to securitize a loan because of negative information about the borrower, rather than for diversification or capital relief reasons (adverse selection due to asymmetric information) complex securitization structures -> investors rely blindly inaccurate credit ratings easy securitization has increased the total amount of lending creating additional demand for housing and pushing housing prices even further away from their fundamental value many originators invested in asset-backed securities by creating structured investment vehicles (SIV) = virtual banks investing in asset-backed securities and funding themselves using short-term debt Tranching prepayment is unattractive to investors (want fixed-income securities with predictable payments) collateralized mortgage obligations (CMOs) = securities that redirect the cash flows of an MBS pool to various segments alternatively: collateralized bond obligations (CBOs), collateralized loan obligations (CLOs), collateralized debt obligations (CDOs) Seite 21 von 64

Tranching = rearrangement of the total cash-flows, total value and total risk of the underlying securities -> nevertheless cash-flows and risks are fully preserved (only redistributed same market value and risk profile) better tailored to investors needs sequential-pay tranches = prioritizing the payment of principal into different tranches planned amortization class (PAC) = to minimize prepayment risk which is transferred to support bonds -> PAC bond offer a fixed redemption schedule as long as prepayments on the collateral stays within a specific range = PAC collar IO/PO-structure: strips the MBS into two components - interest-only (IO): receives only the interest payment on the underlying MBS -> sum of all interest payments depends on the timing of principal payments interest rates fall = faster prepayment = less interest payments over the life of the MBS = IO depreciates in value interest rates rise = slower prepayments = more interest payments over the life of the MBS = IO appreciates in value -> are bullish securities with negative duration - principal-only (PO): receives only the principal payment on the underlying MBS -> sum of all payments are constant but the timing is uncertain interest rates fall = contraction risk interest rates rise = extension risk = PO loses value (principal is paid later, discount rate increases) 8 Fixed-Income Derivatives fixed-income derivatives = instruments whose value derives from a bond price, interest rate or other bond market variable Forward Contracts Forward rate agreements (FRAs) = over-the-counter financial contracts that allow counterparties to lock in an interest rate starting at a future time (buyer = borrowing rate -> benefits from an increase in rates, seller = lending rate -> benefits from a fall in rates) Notation: 1 x 4 (settlich in one month on three-month LIBOR) -> 1 first settlement date, 4 time to final maturity on the settlement date (in one month) the payment to the long involves the net value of the difference between the spot rate and of the locked-in forward rate -> settled in cash, no upfront investment VT = (ST F) * * Notional * PV($1)
ST spot rate F locked-in forward rate the period to which the Index applies PV($1) = $1/ (1 + ST )

a short position in an FRA = equivalent to borrowing short-term to finance a long-term investment -> similar to a long position in a bond duration = difference between the durations of the two legs = (for a short FRA)

Seite 22 von 64

Dollar Duration (DD) = * Notional * PV($1) Futures traded on organized exchanges Eurodollar Futures tied to a forward LIBOR rate (e.g. Euribor futures denominated in Euro, Euroyen futures denominated in Japanese yen) are akin to FRAs involving three-months forward rates starting on a wide range of dates, up to 10 years into the future -> settled in cash, constant duration = 3 month Value of one contract Pt = 10,000 * [100 * 0.25(100-FQt)] = 10,000 * [100 * 0.25 Ft] Settlement value Pt = 10,000 * [100 * 0.25 St]
FQt quoted Eurodollar futures price Ft interest rate 0.25 three month maturity St spot rate

increase in interest leads to a decrease in price of the contract (negative correlation, moves like a bond) -> funds have to be provided precisely when the borrowing cost or reinvestment rate is higher convexity adjustment: adjustment of the futures rate/forward rate due to the increased volatility of marked-to-market transactions Futures rate = Forward rate + (1/2) t1 t2
2

volatility of the change in the short-term rate t1 time to maturity of the futures contract t2maturity of the rate underlying the futures contract

T-Bond-Futures tied to a pool of Treasury bonds that consists all bonds with a remaining maturity greater than 15 years (noncallable) -> similar contracts on shorter rates (2-, 5-, 10-years Treasury notes) settled by physical delivery -> conversion factor (CF) to ensure interchangeability between the deliverable bonds multiplies the futures price for payments to the short, attempts to equalize the net cost of delivering the various eligible bonds minimizes the possibility of market squeezes = when holders of the short position cannot acquire or borrow the securities required under the terms of contract high coupon bonds have higher CFs is not perfect when the term structure is not flat or when the bonds do not trade at their theoretical prices short position delivers bond, receives futures price times conversion factor specific to the delivered bond (set by the exchange at initiation of the contract for each bond) + accrued interest Cost = Price Futures Quote * CF cheapest-to-deliver (CTD) = bond with the lowest net cost the longer the maturity, the greater the difference (lower CF) -> net cost is greater for long-term bonds -> favours short-term bonds for delivery equilibrium futures price: Fte-rt = St PV(D)
St gross price of the CTD PV(D) present value of the coupon payments

Seite 23 von 64

 duration of the futures contract is given by that of the CTD -> approximation as the short only has the option to deliver the CTD -> option should depress the futures price (long is short the option and therefore requires a lower acquisition cost) Swaps = agreements by two parties to exchange cash flows in the future according to a prearranged formula interest rate swaps: payments tied to an interest rate - fixed-for-floating swap: one party commits to pay a fixed percentage of notional against a receipt that indexed to a floating rate (typically LIBOR) -> risk = change in the level of rates - basis swaps: both payments are indexed by a floating rate -> risk = change in the spread between reference rates Instruments in terms of actual cash flows, swap payments are typically netted against each other

Quotation quoted in terms of spreads relative to the yield of similar-maturity Treasury notes e.g. 10 year swap spread as 31/34 bp against LIBOR -> 3 bp spread - dealer is willing to pay: current yield + 0.31 against receiving LIBOR - dealer is willing to receive: current yield + 0.34 against paying LIBOR swap should trade at a positive credit spread to Treasuries interest rate swap rate market is by far the largest derivative market in terms of notional -> very liquid market, market quotations for the fixed leg have become benchmark interest rates basis for the swap curve/ par curve (is equivalent to yields on bonds selling at par) -> floating leg indexed to LIBOR carries credit risk -> swap curve is normally higher than the par curve for government bonds in the same currency Pricing either by taking the difference between two bond prices or valuing a sequence of forward contracts swap is equivalent to a long position in a fixed-rate bond with similar coupon characteristics and maturity offset by a short position in a floating rate note (FRN) value of the swap (V) = BF Bf
BF value of the fixed rate bond Bf value of the floating rate note -> should be close to par

just before reset: the value of the FRN behaves exactly like a cash investment, as the coupon for the next period will be set to the prevailing interest rate -> market value is close to Seite 24 von 64

face value -> just after reset: the FRN behaves like a bond with a six-month maturity fluctuations in the market value of the FRN should be small as difference between two bond prices: - at initiation the swap coupon is set to the prevailing par yield -> Bf = face value (= 100) swap value = 0 -> is like a forward contract at initiation - after the swap is consummated, the value will be affected by interest rates (interest rates fall: swap will move in-the-money -> receives higher coupons than prevailing market yields) -> BF will increase, Bf will barely change - duration of a receive-fixed swap is similar to that of a fixed-rate bind, including the fixed coupons and principal (included in duration calculation even though it is not exchanged!) at maturity (duration of the floating leg is close to 0) as sequence of forward contracts: - value of long forward contract (benefits of rates going up)
ni notional amount for maturity i Fi current forward rate K prespecified rate Ri spot rate (discrete compounding) i time

- swap: receives a fixed-rate -> looses money if rates go up -> sign of the equation is reversed interest rate swaps can be priced and hedged using a sequence of forward rates -> in practice the practice of daily marking-to-market futures induces a slight convexity bias in future rates, which have to be adjusted downward to get forward rates

Options large variety of fixed-income options, often embedded in many securities underlying can be a yield or a price Caps and Floors Cap = call option on interest rates with value:
K = iC cap rate iT rate prevailing at maturity

are purchased jointly with the issuance of floating-rate notes that pay LIBOR plus a spread on a periodic basis for the term of the note -> issuer ensures that the cost of capital will not exceed the capped rate such caps are really a combination of individual options = caplets payment on each caplet is determined by the CT, the notional, and an accrual factor -> payments are made in arrears (= at the end of the period) typically price using a variant of the Black model, assuming that interest rate changes are lognormal -> value of the cap is set equal to a portfolio of N caplets (= European-style individual options on different interest rates with regularly spaced maturities)
ci price of a caplet F current forward rate for period ti to ti+1 K current cap rate PV($1) discount factor to time ti+1

Seite 25 von 64

 volatility is that of the forward rate between now and the expiration of the option contract (ti) -> quoted as one number for all caplets within a cap = flat volatilities (see spot volatilities = quotes separately for each forward rate in the caplet) cap is at-the-money, if the cap rate is the same as the prevailing swap rate (near caplets are out-of-the money, distant caplets are in-the-money) Floor = put option on interest rate with value:
K = if floor rate iT rate prevailing at maturity

Collar = a combination of buying a cap and selling a floor decreases the net cost of purchasing the cap protection when the cap and floor rates converge to the same value (K = iC = iF), the overall debt cost becomes fixed instead of floating -> collar is the same as a pay-fixed swap, which is the equivalent of put-call-parity Long Cap (iC = K) + Short Floor (iF = K) = Long Pay-Foxed Swap Swaptions = OTC options that give the buyer the right to enter a swap at a fixed point in time at specified terms, including a fixed coupon rate European swaption = exercisable on a single date at some point in the future -> owner has the right to enter a swap with a specific rate and term (e.g. 1Yx5Y swaption = in one year a long or short position in a 5 year swap) European 6/1 put swaption or one into five-year payer option: company wants to swap the floating payments (= from issuing a five-year floating rate debt in one year) into fixed payments -> purchase a swaption that will give it the right to create a five-year pay-fixed swap -> if the prevailing swap rate is higher than the agreed rate = exercise the swaption equivalent to a one-year put option on a six-year bond (creates a profit if rates rise and therefore bond value falls) Value of the option at expiration: iT prevailing swap rate for the swap maturity
K locked-in swap rate V(i) value of a swap to pay a fixed rate i

cf. swaption that gives the right to receive fixed is akin to a call option on a bond used for various purposes are typically priced using a variant of the Black model, assuming that interest rates are lognormal -> value of the swaption is equal to a portfolio of options on different interest rate all with the same maturity are traded in terms of volatilities instead of option premiums (in practice) -> applicable forward rate starts at the same time as the option, with a term equal to that of the option Exchange-Trade Options options on Eurodollar futures and on T-bond futures Options on Eurodollar futures give the owner the right to enter a long or short position in Eurodollar futures at a fixed price Seite 26 von 64

 payoff put option:

K strike price FQT prevailing futures price quote at maturity

-> in addition to the cash payoff the option holder enters a position in the underlying futures = put, which creates a short position after exercise with the counterparty taking the opposite position -> futures are settled daily, value of contract = 0 -> payoff is equivalent to that of a cap on rates: Eurodollar future equivalent a caplet on LIBOR with the difference that options on Eurodollar futures are American style and that payments are paid at the expiration date of the Eurodollar futures options instead of in arrears Options on T-bond futures give the owner a right to enter a long or short position in futures at a fixed price payoff on a call option investors who thinks that rates will fall/ bond market will rally = buy a call on Tbond futures -> participate in the upside, without the downside risk

Product Fixed/Floating Swap Cap Floor Put Swaption (payer option) Cal Swaption (receiver option)

Long (buy) Pay fixed Receive floating Pay premium Receive Max(i ic,0) Pay premium Receive Max(iF i,0) Pay premium Option to pay fixed and receive floating Pay premium Option to pay floating and receive fixed

Short (sell) Pay floating Receive fixed Pay Max(i ic,0) Receive premium Pay Max(iF i,0) Receive premium Receive premium If exercised: pay floating and receive fixed Receive premium If exercised: pay fixed and receive floating

9 Equity, Currency, and Commodity Markets

Common stocks/Equities = securities that represent ownership in a corporation bonds are senior to equities (= have a prior claim to the firms assets in case of bankruptcy) equities represent residual claims to what is left of the value of the firm after bonds, loans and other contractual obligations have been paid off limited liability: the most shareholders can lose is their original investment (cf. owners of unincorporated businesses: creditors have claim on personal assets) preferred stocks: differ from common stocks because they promise to pay a specific stream of dividends

Seite 27 von 64

-> behave like a perpetual bond/consol -> but failure to pay the dividends does not result in default -> corporations must withhold dividends to common stock holders until the preferred dividends (cumulative preferred dividends = all current and previous postponed dividends) have been paid out (preferred stocks are junior to bonds but senior to common stocks) -> usually have no voting rights -> are not tax-deductible expenses but have an offsetting tax advantage (corporations receiving preferred dividends only pay taxes on 30% of the amount received -> lowers income tax burden) -> market capitalization and trading volumes is much lower than that of common stocks Valuation extremely difficult value derives from future benefits (stream of future cash flows or future stock price) Gordon-growth model
D dividend r constant discount rate g growth rate

firms value can be assessed using the net present value formula like a bond, as long as the discount rate > growth rate -> growth rate of dividends is uncertain and the required discount rate is not clear, dividends are not always paid out large variations in equity prices can arise from small changes in the discount rate or in the growth rate of dividends -> large volatility of equities risk and expected return of the equity depends on the underlying business fundamentals as well as on the amount of leverage/debt in the capital structure for financial intermediaries the value of underlying assets can be measured precisely -> value the equity from the underlying assets and the cost of borrowing -> more akin to the pricing of a derivative from the price of the underlying Convertible Bonds and Warrants have option like features exercise of convertible bonds/warrants entails the creation of new shares as the option is sold by the corporation itself -> existing shares are said to be diluted by the creation of new shares Warrants = long-term call options issued by a corporation on its own stock -> typically created at the time of a bond issue -> trade separately from the bond to which they were originally attached -> exercised = cash inflow to the firm which issues more shares Convertible Bonds = bonds issued by a corporation that can be converted into equity at certain times using a predetermined exchange ratio -> are equivalent to a regular bond plus a warrant allows the company to issue debt with a lower coupon than otherwise conversion rate = rate with which the conversion can take place (e.g. conversion rare 10 = one bond converted into 10 shares) conversion price = strike price of the option = bond price / conversion rate conversion value = current stock price time the conversion ratio -> convertible bond value must be higher than the price of an otherwise identical straight bond and the conversion value Seite 28 von 64

-> high value of stock price = default of firm unlikely = straight bond price constant (reflecting the discounting of cash flows at the risk free rate) -> option is certain to be exercised: convertible value = close to conversion value -> low value of stock price = default of firm likely = drop of straight bond price (reflecting the loss upon default) -> option will most certainly not be exercised: convertible value = close to straight bond price corporation will typically issue the convertible deep out-of-the-money convertible is also sensitive to interest rate risk Valuation: Warrants can be valued by adapting standard option price models to th dilution effect of new shares Value at origin: N Outstanding Shares V0 = NS0 + MW0
M Outstanding Warrants S0 Initial Stock Price W0 up-front value of Warrant

After dilution: WT = MAX(ST-K,0) = (ST-K) = ((VT +MK)/(N+M) K) -> > 0 WT = ((VT NK ) / (N+M)) = /(N+M) *((VT NK ) = N/(N+M)*( VT/N-K) = equivalent to n = N/(N+M) Warrant can be valued by standard option model with asset price equal to the stock price plus the warrant proceeds, multiplied by n with the usual parameters W0 = n* c (S0 + M/N W0, K, , , r, d) then unit asset value = V0/N = S0 +M/N*W0 -> this must be solved iteratively => W0 on both sides , if M =is small to current float , or number of outstanding shares N W0 = c (S0 + M/N W0, K, , , r, d) On complication is that most convertibles are also callable at the discretion of the firm convertible securities called for several reasons -> First: Issue can be called to force conversion into common stocks when stock price is high enough -> bondholders have typically a month during which they can still convert -> its a forced conversion -> this call feature gives corporation more control over conversion -> allows company to raise equity capital by forcing bondholders to pay exercise price Second: The call may be exercised when option value is worthless and firm can refinance its debt at lower coupon -> similar to a call of nonconvertible bond -> except convertible must be busted-> occurs if stock price much lower than conversion price

Seite 29 von 64

In term of risk factors, a long position in a convertible bond is exposed to increasing interest rates and credit spreads -> e.g. regular corporate bonds but also to factors that decrease the value of the embedded call -> e.g. decreasing stock price and decreasing implied volatility. Equity Derivatives Stock Index Futures Are actively traded all over the world -> turnover corresponding to the notional amount is often greater as amount of trading in physical stocks in same market-> success is explainable by their versatility for risk management -> they allow investors to manage efficiently their exposure to broad stock market movements -> speculators can take directional bets with futures on the upside or downside-> hedgers find that futures provide cost-efficient method to protect against price risk most active contract = S&P 500 future contract on Chicago Mercantile Exchange (CME) -> notional is defined as $250 times index level contracts are cash settled -> do not involve delivery of underlying stocks at expiration -> future contract is priced according to the usual cash-and-carry relationship Fte-rT = Ste-yT
y dividend vield defined by unit of time

Single Stock Futures In late 2000 USA passed legislation authorizing trading in single stock futures -> are future contracts on individual stocks-> Europe and elsewhere already trading -> USA started electronic trading in November 2002 -> each contract gives obligation to buy or sell 100 shares of the underlying -> settlement usually involves physical delivery -> have many advantages-> positions can be established more efficiently due to their low margin requirements, generally 20% of cash value -> also short selling eliminates the costs of inefficiencies associated with stock loan process-> other than physical settlement these contracts trade like stock index futures Equity Options Options can be traded on individual stocks, on stock indices, or on stock index futures. In USA stock options trade on Chicago Board Options Exchange (CBOE) -> Each option gives right to buy or sell a round lot of 100 shares -> settlement involves physical delivery-> traded options are typically American-style their valuation should include possibility of early exercise -> in practise value do not differ from European-style option -> when stock pays no dividend values are same for more precision use of numerical models such as binominal tree -> in USA most active index option is S&P 100 and S&P 500 index traded on CBOE S&P 100 is American-style , S&P 500 is European-style -> options are cash settled -> each contract is for $100 times the value of index-> European options on stocks can be priced using Black-Scholes formula, using y as the dividend yield on the index -> options on S&P 500 stock index also very popular-> give right to enter long or short futures position at a fixed price -> exercise is cash settled Equity Swaps Is an agreement to exchange cash flows tied to return on a stock market index exchange for a fixed or floating rate of interest-> e.g. swap that provides the return of S&P 500 index every six months in exchange for payment of LIBOR plus spread ->swap will be typically priced -> so as to have zero value at initiation-> equity swaps can be valued as portfolios of forward contracts, as in case of interest swaps -> pricing method of currency swaps can also be used > are used by investment managers to acquire exposure to, e,g, an emerging stock market without to invest in the market itself -> can be used to skirt restrictions on foreign investments Seite 30 von 64

Variance Swaps Is forward contract on variance -> payoff: = 252/[ln(Si/Si-1)]

VT = (t0,T KV) N

N national Amount realized variance over life of contract KV strike price or forward price

Can be written on underlying asset -> but most common for equities or equity indices-> allow trades based on direct views on variance -> long positions are bets on high volatility -> initial value of contract is zero -> the widely quoted VIX index is fair strike price for a variance swap on S&P 500 index, quoted as volatility -> market value for outstanding variance swap with = T t days remaining to maturity is Vt= Ne [w(t0,t KV) + (1 w) ( Kt-KV)]
-r

t0,t elapsed variance beween initial time t0 and curren time r w is fraction of days elapsed since t0 Kt is current forward price

contracrs allow also correlation trading -> e.g. index of two stocks -> variance swap is available for each constituent stock as well as for index -> realized variance depends on two variances and the correlation coefficient -> all else equal -> higher correlation = higher portfolivariance -> long correlation trade would buy a variance swap on the index and short variance swaps on components -> if correlation increases long position should gain more than short positions

Currency Markets =forex -> enormous trading activity -> spot transactions are exchanges of two currencies for settlement as soon as it is practical -> typically two business days -> ca 35% of trading volume Outright forward contracts = agreements to exchange two currencies at future date -> ca 12% Forex swaps = involve two transactions, an exchange of currencies on given date and a reversal at a later date -> 35% -> typically short-term nature -> do not confuse with long-term currency swaps = involve stream of payments over longer horizons in addition market also includes OTC forex options and exchanged-traded derivates -> Most active currency futures traded on CME -> settled by physical delivery -> CME also trades options on currency futures -> currency forwards, futures, and options can be priced according to standard valuation models-> specify income payment as continuous flow Currencies are generally quoted in European terms = in units of foreign currency per dollar -> Two exceptions: GBP, EUR -> quoted in American terms = dollars per unit of foreign currency Currency Products currency markets offer full range of financial instruments -> one type of instrument is specific -> quanto = quantity-adjusted derivate -> payoff defined by variables associated with one currency but is paid in another currency -> e.g. U.S investor consider buying Japanese stocks represented by Nikkei 225 index -> normally stock buying involves a position in stocks and in Japanese yen currency -> could have situation where Nikkei index increases in value but gain is wiped out by fall in value of yen -> to avoid currency risk investor could buy quanto forward contract on index with expiry in one year -> contract has notional in dollars and Forward price in yen -> payout at expiration depends on the yen value of index VT =NUSD(ST-FQ) forward price of quanto depends on usual forward price in foreign currency but also the covariance between movements in the yen stock price and in the dollar price of the yen Seite 31 von 64

FQ=F exp(pSTFX) Currency Swaps Are agreements by two parties to exchange a stream of cash flows in different currencies according to prearranged formula Instruments Consider two counterparties, A and B that can raise funds either in dollars or in yen -> A want to raise dollars and B wants to raise yen -> company A has absolute advantage in the two markets -> B has comparative advantage in raising dollars -> provides basis for swap -> if both raise funds in desired markets cost can be higher as than both companies raise funds in the market their they have comparative market -> both parties benefit from swap -> payments are typically netted for interest rate swap -> because currency is same -> not the case for currency swaps -> full interest payments are made in different currencies -> in addition at initiation and termination there is exchange of principal in different currencies Pricing As with interest rate swaps -> swaps can be priced using either of two approaches -> taking the difference between two bonds or valuing a sequence of forward contracts -> swap is equivalent to a long position in a fixed-rate and a short-position -> value of swap is that of long yen bond minus dollar bond V = S($/Y)P*(Y)-P($) Initial value of swap is zero -> assuming flat term structure for both countries and no credit risk Three conditions under which swap will be in the money If value of yen appreciates If yen interest rate falls If dollar interest rate goes up Swap exposed to three risk-factors: spot-rate, and two interest rates -> the latter exposures are given by the duration of equivalent bond A position in a receive-foreign currency swap is equivalent to a long position in foreign currency bond offset by a short position in a dollar bond Swap can be alternatively valued as a sequence of forward contracts Vi= (Fi-K)exp(-rii)
ri dollar interest rate Fi prevailing forward rate in $/yen K locked in rate of exchange

Extending to multiple maturities V = ni(Fi-K)exp(-rii) Commodities Products Commodities are typically traded on exchanges -> contracts include spot, futures and options on futures-> also OTC market for long-term commodity swaps -> payments are tied to price of a commodity against a fixed or floating rate Commodity contracts can be classified into Agricultural products ( including grains and oilseeds (corn, wheat, soybean), food and fiber (cocoa, coffee, sugar, orange juice) Livestock and meat (cattle, hogs) Seite 32 von 64

 Base metals (aluminium, copper, nickel, and zinc) Precious metals (gold, silver, platinum) Energy products (natural gas, heating oil, unleaded gasoline, crude oil) Goldmann Sachs Commodity Index (GSCI) is broad production-weighted index of commodity price performance -> composed of 24 liquid exchange-traded futures contracts -> contains 72% energy products, 7% industrial metals, 2% precious metals, 14% agricultural products, and 5% livestock products -> CME trades futures and option contracts on GSCI -> few last years markets developed for electricity products -> more recently weather derivates Pricing of Futures Commodities differ from financial assets in two notable dimensions May be expensive or impossible to store May generate a flow of benefits that are not directly measureable First dimension involves cost of carrying physical inventory of commodities -> for most financial instruments this cost is negligible -> for bulky commodities this cost may be high -> other commodity cannot be stored easily Second involves benefit from holding the physical commodity -> e.g. company that manufactures copper pipes benefits from an inventory of copper -> this flow is also called convenience Yield for holder -> for financial asset this flow would be a monetary income payment for investor -> when asset can be lent out for profit the yield represents lease rate -> Consider storage cost only -> cash and carry relationship should be modified -> compare two positions -> in the first we buy the commodity spot plus a pay up-front the present value of storage costs -> in the second -> enter forward contract and invest the present value of the forward price -> since two positions are indentical at expiration -> must have same initial value Ft e-r = S t+ PV[C]
e-r present value factor

Alternatively -> storage costs are incurred per unit of time and defined as c Ft e-r = S tec Due to these costs forward rate should be much greater than spot rate Convenience yield can be expressed as y per unit of time -> y represents the net benefit from holding the commodity, after storage costs Ft e-r = S te-y
e-y actualization factor

Actualization factor may have economically identifiable meaning -> reflecting demand and supply conditions in the cash and futures markets -> can be viewed as plug-in Future prices are generally increasing with maturity -> reflecting time value of money, storage costs and low convenience yields -> some irregularities reflecting anticipated imbalances between demand and supply Futures and Expected Spot Prices Interesting issue is whether todays futures price gives best forecast of future spot price -> if so it satisfies the expectations hypothesis Ft = Et[ST]

Seite 33 von 64

The reason this relationship may hold is as follows: One-year oil future price F = $20.47 -> if market forecasts that oil prices in one year will be at $25, one could make profit by going long a futures contract at the cheap future price, waiting a year, then buying oil for cheap $20.47 and reselling at the higher price of $ 25 > deviations from this relationship imply speculative profits To be sure, these profits are not-risk free -> they may represent some compensation for risk -> e.g. market is dominated by producers who want to hedge by selling oil futures -> F will be abnormally low compared with expectations -> relationship between futures prices and expected spot prices can be complex For financial assets for which arbitrage between cash and futures is easy -> the future or forward rate is solely determined by the cash-and-carry relationship For commodities the arbitrage may not be so easy -> future prices may deviate from the cash-and carry relationship through this convenience yield factor -> such prices reflect expectations of futures spot prices, as well as speculative and hedging pressures Market is said to be in contango when futures price trades at a premium relative to spot price Market is said to be in backwardation when forward prices trade at discount relative to spot price With backwardation the future price trends to increase as the contracts near maturity -> in such a situation a roll-over strategy could be profitable -> roll-over strategy involves buying a long maturity contract, waiting, and then selling it at a higher price in exchange for buying a cheaper longer-term contract Strategy is comparable to riding the yield curve when upward-sloping = involves buying long maturities and waiting to have yields fall due to the passage of time Markets are in contango if spot prices are lower than forward prices. Markets are in backwardation if spot prices are higher than forward prices. Backwardation occurs when there is a high current demand for the commodity, which implies high convenience yields

10 Introduction to Market Risk

market risk is primarily measured with position-based risk measures (e.g. value-at-risk) Value-at-risk (VAR) = a statistical measure of total portfolio risk, based on the most current positions, which takes into account portfolio diversification and leverage risk managers should consider the entire distribution of profits and losses over the specified horizon -> in practice, the distribution is summarized by the worst loss at a specified confidence level stress-testing = identifies potential losses under extreme market conditions Types of Financial Risks financial risk = market risk, credit risk, operational risk market risk = the risk of losses due to movements in financial market prices or volatilities -> usually includes liquidity risk = risk of losses due to the need to liquidate positions to meet funding requirements -> not amendable to formal quantification credit risk = the risk of losses due to the fact that counterparties may be unwilling or unable to fulfil their contractual obligations -> settlement risk (=the risk that a counterparty does not deliver a security or its value in cash as per agreement when the security was traded after the other counterparty or counterparties have already delivered security or cash value as per the trade agreement) / Herstatt risk (= cross-currency settlement risk that arises where the working hours of inter-bank fund transfer systems do not overlap due to time zone differences) Seite 34 von 64

 operational risk = the risk of loss resulting from failed or inadequate internal processes, systems and people or from external events -> interact with each other Risk management tools variety of ad hoc tools: notional amounts, sensitivity measures, scenarios -> do not measure the potential for downside loss for the total portfolio -> fail to take into account differences in volatilities across markets, correlations across risk factors, as well as the probability of adverse moves in the risk factors -> notional amount: only provides an indication of the potential loss -> sensitivity measure e.g. duration of a bond: reveals the extreme sensitivity of the bond to interest rates, does not reveal if a disastrous movement is likely, ignores the nonlinearity between note price and yields -> scenario analysis: allows the investor to investigate nonlinear, extreme effects in price -> method does not associate the loss with a probability -> do not allow the investor to aggregate risk across different markets VAR = aggregates risks across the whole portfolio, taking into account leverage and diversification and providing a risk measure with an associated probability -> was endorsed by the G-30 as part of best practice for dealing with derivatives (1993) -> result from a merging of finance theory (focus on pricing and sensitivity of financial instruments) and statistics (studies the behaviour of the risk factors) VAR = Market value * Modified Duration * Worst yield increase measures such as notional amounts and exposures are still used to set limits (= attempt to control risk before it occurs) stop-losses = rules enforcing position cuts after losses occur -> only provide partial protection because they are applied after a loss Downside risk measures VAR = a summary measure of downside risk expressed in dollars or in the reference currency = the maximum loss over a target horizon such that there is a low, prespecified probability that the actual loss will be larger -> takes into account the time value of money -> is typically reported as a positive number even if it is a loss -> the time horizon is usually very short in which case the average return on financial series is close to zero is sometimes reported as the deviation between the mean and the quantile confidence level c (e.g. 95%) -> corresponds to a right-tail probability confidence level p = 1 c -> left-tail probability verify the quality of VAR forecasts by checking whether the exceptions (= number of losses worse than VAR) is in line with expectations number of expected exceedences n = p * N
N period of days P left-tail probability

under the null hypothesis that the model is correctly calibrated or that the true probability of exceptions is p, the distribution of number of exceptions x follows a binomial distribution

Seite 35 von 64

 distribution can be used to compute a cutoff value for the number of exceptions, beyond which it is concluded that the model is flawed exception tests only focus on the frequency of occurrences of exceptions -> do not take into account the size of losses VAR caveats: does not describe the worst loss: expect the VAR number to be exceeded with a frequency of p does not describe the losses in the left tail: does not say anything about the distribution of losses in its left tail -> indicates just the probability of such a value occurring -> very different distribution shapes possible for the same VAR number -> fails to qualify as a subadditive risk measure: VAR of a portfolio could be greater than the sum of the VARs for its components -> subadditivity = the risk of a portfolio must be less than the sum of risk for portfolio components is measured with some error: subject to normal sampling variation -> other sample period/ different length will lead to a different VAR number past window should reflect the range of future outcomes Alternative Measures of Risk the entire distribution: VAR is simply one quantile in the distribution -> risk manager has access to the whole distribution and could report a range of VAR numbers for increasing confidence levels the conditional VAR (CVAR) = expected value of the loss when it exceeds VAR also called: expected shortfall, tail conditional expectation, conditional loss, expected tail loss -> is qualified as a subadditive risk measure (because it is an average of the tail loss)
- q VAR number

the standard deviation: takes into account all observations -> can be translated into a VAR measure using a multiplier that depends on the distribution and the selected confidence level

volatility of the rate of return W amount invested measured in the reference currency multiplier c confidence level expected rate of return over the horizon

measures VAR relative to the mean (standard deviation is a measure of dispersion around the mean) -> can be measured relative to the initial value Seite 36 von 64

 disadvantage: it is symmetrical and cannot distinguish between large losses or gains -> requires a distributional assumption which may not be valid the semistandard deviation: simple extension of the usual standard deviation that considers only point data that represent a loss
NL number of data points that represent a loss

accounts for asymmetries in the distribution (e.g. negative skewness) sometimes used to report downside risk but is much less intuitive and less popular the drawdown: = a decline over a fixed time interval
xMAX local maximum over the period which occurs at time t MAX [0,T] t time

maximum drawdown is the largest such value over the period, or decline from peak to trough (local maximum to local minimum) is useful if returns are not independent from period to period useful measures of risk if the portfolio is actively managed disadvantage: backward-looking -> cannot be constructed from the current position -> maximum drawdown measures are not directly comparable across portfolios (different time intervals) Cash Flow at Risk VAR methods to measure the mark-to-market risk of commercial bank portfolios Cash Flow at Risk (CFaR) measures the worst shortfall in cash flows due to unfavourable movements in market risk factors CF = Q * (P C)
Q Quantities P unit revenues C unit costs

risk manager needs to model the relationship between quantities and risk factors -> simulations can be used to project cas-flow distribution and identify the worst loss at some confidence level over the long-term, the firm may have strategic options (e.g. withdraw from a foreign market, open up an assembly operation in the foreign country, etc.) -> incorporated in the risk analysis VAR Parameters Confidence Level the higher the confidence level, the greater the VAR measure -> provides useful information about the return distribution and potential extreme losses increasing the confidence level will create an increasingly larger loss but less likely -> number of occurrences below VAR shrinks leading to poor measures of high quantiles choice depends on the use of VAR -> if used as a benchmark measure of downside risk, what matters is the consistency of the VAR confidence level across trading desks or time -> if Seite 37 von 64

used to decide how much capital to set aside to avoid bankruptcy, a high level is advisable (capital adequacy use -> applied to the overall institution) backtesting systematically checks whether the frequency of losses exceeding VAR is in line with p = 1-c -> too high confidence level will make it impossible to verify the true probability associated with VAR Horizon the longer the horizon, the greater the VAR measure -> two factors: behaviour of the risk factors and the portfolio positions# to extrapolate from a one-day-horizon to a longer horizon, it is assumed that returns are independently and identically distributed -> daily volatility can be transformed into a multiday volatility (multiplication by the square root of time valid with independent and identically distributed returns that have a normal distribution) -> assumed that the distribution of daily returns is unchanged for longer horizons

requires the distribution to be invariant to the horizon, the distribution to be the same for various horizons, and innovations to be independent across days choice depends on the use of VAR -> if used to provide an accurate benchmark measure of downside risk, the horizon should be relatively short (ideally < than the overage period for major portfolio rebalancing) -> if used to decide how much capital to set aside to avoid bankruptcy, a long horizon is advisable in practice: cannot be less than the frequency of reporting of profits and losses (usually banks at a daily basis, corporates on a longer interval) = minimum horizon for VAR shorter time intervals create more data points matching the forecast VAR with the actual, subsequent P&L -> power of statistical tests increases with the number of observations needs to be appropriate to the asset classes and the purpose of risk management The Basel Rules Basel Committee on Banking Supervision: minimum capital requirements for commercial banks to cover market risk Market Risk Charge (MRC) based on VAR models with the following parameters: horizon of 10 trading days or 2 calendar weeks confidence level 99% observation period based on at least a year of historical data and updated at least once a quarter Internal Models Approach (IMA): MRC is computed as the sum of a general market risk charge (GMRC) plus a specific market risk charge (SRC) -> GMRC involves the average of the trading VAR over the last 60 days, times a supervisordetermined multiplier (minimum value = 3), as well as yesterdays VAR -> SRC = defined to provide a buffer against losses due to idiosyncractic factors related to the individual issuer of the security -> included the risk that an individual debt or equity moves by more or less than the general market factors, reflecting basis risk or event risk such as downgrade or default

Components of Risk Measurement Systems Seite 38 von 64

 Steps of a risk measurement system construct the distribution of risk factors from market data collect the portfolio positions and map them onto the risk factors construct the portfolio VAR using one of the three methods (delta-normal, historical, Monte Carlo)

Portfolio Positions assumption that all positions are constant over the horizon (simplification) true risk can be greater (if VAR is based on close-to-close positions that reflect lower trader limits) or lower (if management enforces loss limits) than the VAR measure Risk Factors represent a subset of all market variables that adequately span the risks of the current or allowed portfolio -> the more complex the strategies, the greater the number of risk factors that should be used VAR Methods choice depends on the nature of the portfolio Stress-Testing aims at identifying situations that could create extraordinary losses for the institution is a key risk management process which includes scenario analysis, stressing models, volatilities and correlations, and developing policy responses scenario analyses: submits the portfolio to large movements in financial market variables > scenarios created using a number of methods moving key variables one at a time: simple and intuitive, but difficult to assess realistic comovements in financial variables using historical scenarios creating prospective scenarios: should be tailored to the portfolio at hand, assessing the worst thing that could happen to the current positions reverse stress-tests : start from assuming a large loss and then explore the conditions that would lead to this loss useful to guard against event risk = the risk of loss due to an observable political or economic event (usually in emerging markets) changes in governments: leading to changes in economic policies changes in economic policies (e.g. default, capital controls, inconvertibility, changes in tax laws, expropriations, etc.) Seite 39 von 64

 coups, civil wars, invasions or other signs of political instability currency devaluation goal of stress-testing = identify areas of potential vulnerability -> to ensure that the institution can withstand likely scenarios without going bankrupt -> can be easily implemented once the VAR structure is in place

11 Hedging Linear RiskSources of Market Risk

downside risk can be decomposed into two types of drivers: exposures (= component which the risk manager can control) and risk factors (= exogenous component) currency risk = the volatility of floating exchange rates and devaluation risk, for fixed currencies fixed-income risk = relates to term-structure risk global interest rate risk, real yield risk, credit spread risk and prepayment risk equity risk = described in terms of country risk, industry risk, stock-specific risk commodity risk = includes volatility risk, convenience yield risk, delivery and liquidity risk Sources of Loss: A Decomposition plain fixed-coupon bond: the potential for loss can be decomposed into the effect of dollar duration D*P and the changes in the yield dy Bond value change dP = - (D*P) * dy losses can occur because of a combination of two factors: the exposure to the factor = dollar duration -> is a choice variable the movement in the factor itself, which is external to the portfolio linear characterization also applies to systematic risk and option delta return can be decomposed into a component due to the market and some residual risk (effect washes out in a large portfolio)
Ri return on stock i -> expressed as rate of return (no dimension) Rm component due to the market i constant i residual

specific risk = risk that is due to issuer-specific price movements, after accounting for general market factors

first partial derivative of the option

change in the value of a derivative can be expressed in terms of the change in the price of the underlying asset change in value is linked to an exposure coefficient and a change in market variable (both necessary for a loss) Market loss = Exposure * Adverse Movement in Financial Variable

Seite 40 von 64

 discontinuities in the portfolio value come from either discontinuous payoffs (arise with some instruments which pay a fixed amount if the option ends in the money and none otherwise) or from jumps in the risk factors (or due to event risk) Currency Risk arises from potential movements in the value of foreign currencies -> includes currencyspecific volatility, correlations across currencies and devaluation risk arises in the following environments in a pure currency float -> external value of a currency is free to move and to depreciate/ appreciate as pushed by market forces (e.g. dollar/euro exchange rate) in a fixed currency system -> currencys external value is pegged to another currency > risk due to possible readjustments in the parity value (devaluation, revaluation) in a change in currency regime -> a currency that was previously fixed becomes flexible and vice versa Currency Volatility across developed markets, volatility typically ranges from 6% to 10% per annum some currencies have very low volatility, reflecting their pegging to the dollar -> devaluation risk = risk that currency peg could fail Correlations correlations between these currencies against the U.S. dollar: low mostly in the range of -0.10 to 0.20 -> indicates substantial benefits from holding a well-diversified currency portfolio blocks of currencies with very high correlations (e.g. European currencies have high correlations with each other and the Euro -> ~ 0,60 to 0.90) -> investing across these currencies does little to diversify risk (from the viewpoint of a U.S. dollar-based investor) Cross-Rate Volatility exchange rates are expressed relative to a base currency (usually the dollar) cross rate = the exchange rate between two currencies other than the reference currency

12 correlation coefficient

assumes that both the nominator and the denominator are in the same currency .> otherwise the log of the cross rate is the sum of the logs Fixed-Income Risk arises from potential movements in the level and volatility of bond yields -> movement of yield curves creates yield curve risk for fixed-income portfolios Factors Affecting Yields movement in yields reflects economic fundamentals determinants of movements in interest rates: inflationary expectations: any perceived increase in the forecast rate of inflation will make bonds with fixed nominal coupons less attractive (= increase in yield) Seite 41 von 64

-> real interest rate = nominal rate rate of inflation over the same period movement in the term structure of interest rates can be summarized by two maturities long-term rate (e.g. from the 10-year Treasury note) short-term rate (e.g. from the 3-month Treasury bill) -> usually move in tandem although the short-term rate displays more variability -> term spread = the difference between the long rate and the short rate -> slow economic activity decreases the demand for capital, which in turn decrease short-term rates and increases term spread -> central banks lower short-term rates to stimulate the economy Bond Price and Yield Volatility risk can be measured as either return volatility or yield volatility

short-term investments have very little price risk due to their short maturity and duration > price risk of 10-year bonds ~ 6% to 10% (similar to that of floating currencies), price risk of 30-year bonds ~ 20% (similar to equity) more intuitive to measure yield volatility (dy) -> remains more constant over time than price volatility, which must approach zero as the bond approaches maturity -> yield volatilites for swaps and zeros ~ 0.50% per annum, typical yield volatility ~ 1% Correlations if the yield curve were to move in strict parallel fashion, all correlations should be equal to 1 -> in practice more complex patterns movements in adjoining maturities are highly correlated -> give risk managers an opportunity to simplify the risk factors by using principal components (= statistical technique that extracts linear combination of the original variables that explain the highest proportion of diagonal components of a matrix) -> level risk factor: explains a high percentage of the total variance and has similar weights on all maturities -> slope risk factor/twist: explains a low percentage of the total variance, is associated with opposite positions on short and long maturities -> curvature risk factor/ bend risk factor/ butterfly Global Interest Rate Risk different fixed-income markets are exposed to their own sources of risk level of yields falls within a remarkable narrow range -> reflects the fact that yields are primarily driven by inflationary expectations, which have become similar across all these markets -> central banks have proven their common determination to keep inflation in check expect the volatility of yield changes to decrease as yields drop toward zero and to be higher when yields are high -> can be modelled by relation the volatility of yield changes to a function o the yield level -> Cox, Ingersoll, Ross model: posits that movements in yields should be proportional to the square root of the yield level correlations are very high across continental European bond markets that are part of the euro -> these markets are now moving in synchronisation (dictated by monetary policy by the ECB) -> Eurozone bonds only differ in terms of credit risk

Seite 42 von 64

Real Yield Risk many countries have issued inflation-protected bonds, which make payments that are fixed in real terms but indexed to the rate of inflation -> source of risk is the real interest rate risk real yield can be viewed as the internal rate of return that will make the discounted value of promised real bond payments equal to the current real price -> movement in real interest rate may not correlate perfectly with movements in nominal yields can be used to infer inflationary expectations -> implied rate of inflation = nominal yield real yield Credit Spread Risk == the risk that yield on duration-matched credit-sensitive bond and Treasury bonds could move differently credit spread represents a compensation for the loss due to default, plus perhaps a risk premium that reflects investor risk aversion a position in a credit spread can be established by investing in credit-sensitive bonds (e.g. corporates, agencies, mortgage-backed securities) and shorting Treasuries with the appropriate duration -> position benefits from a stable or shrinking credit spread, but loses from a widening of spreads distribution of credit spreads is asymmetric (cannot turn negative) -> large move imply increases in spreads -> positions in credit spreads can be expose to large losses Prepayment Risk arises in the context of home mortgages where there is uncertainty about whether the homeowner will refinance his loan early -> prominent feature of mortgage-backed securities: investor has granted the borrower an option to repay the debt early option depends on the age of the loan (seasoning), the current level of interest rates, the previous path of interest rates (burnout), economic activity, seasonal patterns investors can evaluate the attractiveness of MBSs by calculation their option-adjusted spread (OAS) = spread over the equivalent Treasury minus the cost of the option component Equity Risk arises from potential movements in the value of stock prices -> the total risk can be decomposed into a marketwide risk and stock-specific risk Stock Market Volatility risk is much greater than for currencies ( ~12% to 20%) -> markets that are less diversified or are exposed to greater fluctuations in economic fundamentals are more volatile Commodity Risk arises from potential movements in the value of commodity contracts, which include agricultural products, metal and energy products Commodity Volatility commodity contracts of RiskMetrics grouped into precious metals (gold, platinum, silver ~20%-30% volatility), base metals (aluminium, copper, nickel, zinc) and energy products (natural gas, heating oil, unleaded gasoline, crude oil-West Texas Intermediate - ~20%-70% volatility) -> energy products are less storable and are therefore more affected by variations in demand and supply (= higher volatility) Futures Risk
e-rt present factor in the base currency e-yt includes y convenience yield y

Seite 43 von 64

 convenience yield y: represents an implicit flow benefit from holding the commodity -> any storage cost should be deducted from y -> for precious metals the convenience yield is close to zero -> conceptually similar to a dividend yield on a stock index but cannot be measured as regular income -> can be quite volatile future prices are less volatile for longer maturities -> decreasing term structure of volatility is more marked for energy products and less so for base metals movements in futures prices are much less tightly related to spot prices than for financial contracts variations in the basis are much more important for energy products than for financial products, or even metals (lower correlations) spot and future prices for crude oil: much more variation in the basis between the spot and futures prices -> market switches from backwarding (S>F) to contango (S<F) -> futures contracts represent a separate risk factor -> energy risk measurement systems require separate risk factors for each maturity positions in commodity futures are also exposed to delivery and liquidity risk -> asset liquidity risk is due to the relative low volume in some of these markets, relative to other financial products taking delivery or having to deliver on a futures contract that is carried to expiration is costly (transportation, storage, insurance costs) -> requires complying with the type and location of the commodity that is to be delivered Risk Simplification aggregate the portfolio risk at the highest level -> in practice: too complex to model each risk factor individually movements in the terms structure of interest rates could be simplified to a few major risk factors Diagonal Model initially applied to stocks but the methodology can be used in any market starts with a statistical decomposition of the return in stock i into a marketwide return and an idiosyncratic risk adds the assumption that all specific risks are uncorrelated -> any correlation across two stocks come from the joint effect of the market return on stock i: systematic risk / regression slope ratio: residual is uncorrelated with the component due to the market (Rm) equilibrium in capital markets imposes restrictions on the i portfolio that consists of positions wi on the various assets
systematic risk of stock i Rm component due to the market i constant i residual

portfolio return

Seite 44 von 64

portfolio variance if the portfolio is equally weighted and the residual variances are all the same, with increasing N the second term should vanish -> the only remaining risk is the general market risk, consisting of the beta squared times the variance of the market -> justifies ignoring specific risk in large, well-diversified portfolios -> underestimated risk for portfolio with a small number of stocks concentrated in one sector approach can be expanded to multiple factors -> can be used to build a covariance matrix from general market factors -> each security is mapped on the selected risk factors -> exposures are then added up across the entire portfolio, for which risk is aggregated at the top level (useful approach when there is no history of returns for some positions) Fixed-income Portfolio Risk analysis of a corporate bond portfolio with N individual bonds -> simplify the risk profile of the portfolio potential major risk factors are movements in a set of J Treasury zero-coupon rates zj, and in K credit spreads, sk, sorted by credit rating -> goal is to provide a good approximation tot the risk of the portfolio not practical to model the risk of all bonds -> may not have sufficient history or history may not be relevant if it does not account for the probability of default model the movement in each corporate bond yield yi by a movement in the Treasury factor zi at the closest maturity and in the credit rating sk class to which it belongs -> i is assumed to be independent across i

DVBP total dollar value of a basis point for the associated risk factor

portfolio price movement for ni units of this bond:

total variance: if the portfolio is well diversified, the general risk term should dominate -> specific risk (= second term) can be ignored -> portfolio can be characterized by its exposure to just a few government maturities and credit spreads

12 Hedging Linear Risk

risk that has been measured can be managed traditional approach includes hedging -> consists of taking positions that lower the risk profile of the portfolio -> techniques have been developed in the futures market

Seite 45 von 64

-> objective = to find the optimal position in a futures contract that minimizes the variance (more generally the VAR) of the total position -> portfolio consists of two positions: a fixed inventory to be hedged and a hedging instrument -> types of hedging: static hedging: consists of putting on and leaving a position until the hedging horizon dynamic hedging: consists of continuously rebalancing the portfolio to the horizon > can create a risk profile similar to positions in optios hedging creates hedge slippage or basis risk -> basis risk arises when changes in payoffs on the hedge instrument do not perfectly offset changes in the value of the inventory position the objective of hedging is to lower volatility -> will eliminate downside risk but also any upside potential -> objective is to lower risk and not to make profits (trade-off between risk and return) Introduction to futures hedging Unitary hedging amounts transacted are identical in the two markets futures hedging can be quite effective, removing the effect of fluctuations in the risk factor

Q .. amount of foreign currency transacted S spot rates F futures rates -> S and F indexed by 1 at the initial time and by 2 at the exit time

Basis b = S - F profit only depends on the movement of the basis -> effect of hedging is to transform price risk into basis risk a short hedge position is said to be long the basis, since it benefits from an increase in the basis basis risk is minimal if the cash and futures correspond to the same asset, the cash-andcarry relationship holds very well and the remaining maturity is rather short

Basis risk arises when the characteristics of the futures contract differ from those of the underlying position -> futures are standardized to a particular grade but a hedger may have a position in a different grade, which may not be perfectly correlated with the future is the uncertainty whether the cash-futures spread will widen or narrow during the hedging period -> hedging can be effective, if movements in the basis are dominated by movements in cash markets is inevitable for most commodities trade-off between liquidity (increasing of trading volumes by speculators) and basis risk

Seite 46 von 64

 basis risk is higher with cross-hedging -> involves using a futures on a totally different asset or commodity than the cash position basis risk is lowest when the underlying position and the futures correspond to the same asset -> some basis risk remains because of different maturities various time components to hedge using T-bond futures maturity of the underlying bond time to futures expiration hedge horizon -> basis risk occurs when the horizon does not match the time to futures expiration

Optimal hedging decide on the quantity used of the portfolio for hedging -> only risk is price risk if quantity of the inventory is known -> more complex when it involves hedging revenues (analysis of supply and demand conditions necessary e.g. for commodities) The optimal hedge ratio inventory / positions to be hedged can be existing or anticipatory (= to be received in the future with a greater degree of certainty) manager goes long N futures contracts total change in the value of the portfolio
S change in the dollar value of the inventory F change in the dollar value of one futures contract

SF covariance between futures and spot price changes

minimum variance hedge ratio S consists of the number of units (shares, bonds, bushels, gallons) times the unit price (stock price, bond price, wheat price, fuel price) express volatilities in terms of rates of changes in unit prices (are unitless)

optimal hedge ratio

Q quantities s unit prices value of inventory S = Q * s notional amount of one futures contract F = Qf * f sf the coefficient in the regression of s/s and f/f second term represents an adjustment factor for the size of the cash position and of the futures contract

Seite 47 von 64

 best hedge is obtained from a regression of the (change in the ) value of the inventory on the value of the hedge instrument -> is given by the negative of the beta coefficient of a regression of changes in the cash value on changes in the payoff on the hedging instrument

measure the quality of the optimal hedge ratio in terms of the amount by which the variance of the original portfolio decreases (effectiveness of the hedge)

volatility of the hedged position if R2 = 1 , the regression fit is perfect and the resulting portfolio has zero risk (no basis risk) -> if R2 is very low, the hedge is not effective Liquidity issues futures hedging can create other risks -> can involve large cash inflows and outflows (can create liquidity problems, especially when they are not offset by cash inflows from the underlying position) Applications of optimal hedging Duration hedging applies to the bond market modified duration can be viewed as a measure of the exposure of relative changes in prices to movements in yields
D* modified duration D* P dollar duration

assumption that change of the yield does not depend on maturity for the cash and futures positions (including variances and covariances):

D*s / D*f modified durations of S and F

Total portfolio payoff is zero when the net exposure ( value in []) is zero the optimal hedge is simply minus the ratio of the dollar duration of cash relative to the dollar duration of the hedge -> can be expressed in dollar value of the basis point (DVBP) Beta hedging applies to the stock market (equity hedging using stock index futures) Seite 48 von 64

 beta/systematic risk can be viewed as a measure of the exposure of the rate of return on a portfolio i to the movements in the market m
the intercept (is not a source of risk and therefore ignored for risk management purposes) systematic risk residual component

stock-index futures contract which has a beta of unity (F/F) = 1(M/M) -> for options the betas is replaced by the net delta (C) = (M)

-> total portfolio payoff

-> optimal number of contracts: the optimal hedge with stock index futures is given by the beta of the cash position times its value divided by the notional of the futures contract

13 Nonlinear Risk: Options

Evaluating options Definitions Consider a derivate instrument whose value depends on an underlying asset, which can be a price, an index or a rate. E.g. a call option where underlying asset is foreign currency
St current spot price o fasset in dollars Ft current forward price of asset K exercise price of option contract ft current value of derivate instrument rt domestic risk-free rate rt* foreign risk-free rate (als written as y) t annual volatility of the rate of change in S t time to maturity

r* represents the income payment on the asset, which represents the annual rate of dividend or coupon payments on a stock index or bond. For most options, value of derivate as function ft = f(St, rt, rt*, t, K, t) Contract specifications are represented by K and time to maturity t. The other factors are affected by market movements, creating volatility in the value of the derivative. For simplification time subscripts are dropped -> derivatives pricing is all about finding the value of f, given the characteristics of the option at expiration and some assumptions about market behaviour. For a forward contract, e.g., the expression is very simple. f =Se-rt-Ke-rt More generally we may not be able to derive an analytical expression for the function f, requiring numerical methods

Seite 49 von 64

Taylor Expansion: We are interested in describing the movements in f. The exposure profile of derivate can be described locally by taking a Taylor expansion
df =af/asds+0,5af/aSdS+af/ardr+af/ar*dr*+af/ad+af/aTdT+

Because value depends on S in a nonlinear fashion, a quadratic term for S was added Option pricing is about finding f. Option hedging uses the partial derivates. Risk management is about combining those with the movements in the risk factors. Current value of option Delta + gamma estimate Actual price Delta estimate Current price of underlying Graph describes relationship of a European Call and underlying asset. Because we are dealing with sums of local price movements, we can aggregate the sensitives at the portfolio level -> is similar to portfolio duration calculation Defining =af/aS P = xii To hedge against first-order price risk, it is sufficient to hedge the net portfolio delta. = more effective than trying to hedge every single instrument individually. Taylor expansion provides bad approximation in Large movements in the underlying risk factor Highly nonlinear exposures, such as options near expiry or exotic options Cross-partial effects, such as changing in relation with S In such cases we need to turn to a full revaluation of the instrument. Using the subscripts 0 and 1 as the initial and final values, the change in option value is f1-f0= f(S1,r1,r1*,1,K,t1) - f(S0,r0,r0*,0,K,t0) Option pricing: We now present the various partial derivatives for conventional European call and put options. Black Scholes model provides a closed-form solution, from which these derivatives can be computed Key point of BS model is that a position in option can be replicated by a delta position in the underlying asset. a portfolio combining the asset and the option in appropriate proportions is risk-free locally (for small movements in prices) To avoid arbitrage this portfolio must return the risk-free rate. The option value is the discounted expected payoff Ft= ERN[e-rtF(ST)]
ERN expectation of future payoff in a risk-neutral world = underlying asset grows at risk-free rate and the discounting also employs the risk-free rate

In case of European call the final payoff is F(ST) = Max(ST K, 0) and current value of call is given by

Seite 50 von 64

c = Se-r*tN(d1) Ke-rtN(d2)

N(d) cumulative distribution function fort he standard normal distribution defined as standard normal density function

N(d) = (x)dx= 1/2e-1/2xdx

N(d) is also area to the left of a standard normal variable with value equal to d. The values of d1 and d2 are d1= ln(Se-r*t/Ke-rt)/t + t/2 d2 == d1 - t By put-call parity, the European put option value is p = Se-r*t[N(d1)-1] Ke-rt[N(d2)-1] Option Greeks Option Sensitivities: Delta and Gamma Given these closed-form solutions for European options, we can derive all partial derivatives. The most important sensitivity is the delta, which is first partial derivative with respect to the price. For a call option, this can be written explicitly as: c = ac/aS = e-r*tN(d1) Which is always positive and below unity. varies substantially with spot price and with time. As spot price increases, d1 and d2 become very large and tends toward e-r*t, close to one for short maturities in this situation the option behaves like an outright position in the asset If S is very low is close to zero and option is not very sensitive to price. When S is close to strike price is close to 0,5 The delta of an at-the-money call option is close to 0,5. Delta moves to 1 as the call goes deep in-the money. It moves to zero as the call goes deep out-of-the-money The delta of an at-the money put option is close to -0,5. Delta moves to -1 as the put goes deep in-the-money. It moves to zero as the put goes deep out-of-the-money. As option nears maturity -> function becomes more curved -> function converges to step function Lose to maturity options have unstable deltas Gamma is the second-order term which has the bell shape of the normal density function It measures the instability in . Gamma is identical for call and put options with identical characteristics At-the-money options have highest gamma, near maturity the gamma increases For vanilla options, gamma is the highest, or nonlinearities are most pronounced, for shortterm at-the-money options Gamma is similar to concept of convexity developed for bonds Seite 51 von 64

Positive convexity or gamma implies that the value of the asset drops more slowly and increases more quickly than otherwise Long Call : > 0; Gamma > 0 Long Put: < 0; Gamma > 0 Short Call: <0; Gamma <0 Short Put: >0; Gamma < 0

Option Sensitivities: Vega Unlike linear contracts, options are exposed not only to movements in the direction of the spot price, but also in its volatility. Options can be viewed as volatility bets The sensitivity of an option to volatility is called vega, which has also the bell shape of normal density function As with gamma, vega is identical for similar call and put positions Vega must be positive for long positions Vega decreases with maturity Vega is highest for long-term at-the-money options Changes in volatility can be substantial source of risk. Option Sensitivities: Rho = sensitivity to domestic interest rate An increase in the rate of interest increases the value of the call, as the underlying grows at higher rate -> probability of exercising the call increases An increase in dividend yield decreases the growth rate of underlying asset = harmful for value of call Option Sensitivities: Theta Variation in option value due to passage of time is called theta = time decay Unlike other factors, the movement in remaining maturity is perfectly predictable -> time is no risk factor It is generally negative for long positions of calls and puts -> option loses value as time goes by For American options theta is always negative, because of option to exercise early Like gamma, theta is greatest for short-term at-the-money options, when measured in absolute value Option Pricing an the Greeks Having defined the option sensitivities, we can illustrate an alternative approach to the derivation of the Black-Scholes formula. Recall that the underlying process for the asset follows a stochastic process known as geometric Brownian motion (GBM) dS = Sdt + Sdz
dz .. has normal distribution with mean zero and variance dt

Considering only this single source of risk, we can return to taylor Equation. The value of the derivative is a function of S and time. The question is how does f evolve over time? We can relate the stochastic process of f to that of S using Itos lemma. Can be viewed as an extension of the taylor approximation to a stochastic environment. Seite 52 von 64

Applied to GBM: df = (af/asS + af/aS S + af/at) dt + af/aSS)dz df = (S +1/2S + )dt + (S)dz First term including dt is trend, the second, including dz is the stochastic component Next we construct a portfolio delicately balanced between S and f that has no exposure to dz = f -S Its stochastic process is d = (1/2Sn+)dt This simplification is extremely important, the terms involving dz cancel out each other -> portfolio has been immunized against this source of risk. At the same time the terms in S also cancel out each other The fact that disappears from the trend in the portfolio is important, as it explains why the trend of the underlying asset does not appear in the Black-Scholes formula Continuing, we note that the portfolio has no risk. To avoid arbitrage, it must return the risk free rate d = [r]dt = r(f-S)dt If the underlying asset has a dividend yield of y, this must be adjusted to: d =r)dt + ySdt = r(f-S)dt + ySdt Setting the trends equal each other we must have (r-y)S + 1/2S + = rf This is the Black Scholes partial differential equation (PDE), which applies to any contract, or portfolio, that derives its value from S. The solution of this equation with appropriate boundaries conditions, leads to the BS formula for a European call For delta-hedged portfolios, Gamma and Theta must have opposite signs. Portfolios with positive convexity, for example, must experience time decay. Dynamic Hedging BS derivation taught us how to price and hedge options It also showed that holding a call option is equivalent to holding a fraction of the underlying asset, where the fraction dynamically changes over time Delta and Dynamic Hedging Long Call Slope2 Slope1 Seite 53 von 64

Long stock

Graph displays current value of a long call as a function of the current spot price. The long call is replicated by a partial position in the underlying asset. For an at-the-money position the initial delta is 0,5. As the stock price increases from P1 to P2 , the slope of option curve , or delta, increases from 1 to 2 the option can be replicated by larger position in underlying asset. -> if price falls position is cut, as in a graduated stop-loss order dynamic adjustment buys more if asset price goes up and sells it after a fall Dynamic replication of put. Start with delta near -0,5. As price goes up delta increases toward 0 strategy fits long call strategy -> buy when price rises and sell when price falls Short positions in calls and puts show the other pattern -> here buy when falling and sell when rising Implications These patterns are for risk managers very important for the following reasons Dynamic replication of a long position is bound to lose money, because it buys assets after the price has gone up -> each transaction loses small amount of money , which will accumulate precisely to option premium These automatic systems, if applied on a large scale, have potential to be destabilizing -> selling on downturn in price can exacerbate the downside move This pattern of selling asset after its price went down is similar to prudent risk management practices -> typically traders must cut down positions after they incur large losses -> is similar to decreasing delta if S drops loss-limit policies bear some resemblance to a long position in an option Success of this replication strategy critically hinges on the assumption of a continuous GBM price process with this process theoretically rebalance of portfolio as often as needed is possible -> in practice the replication might fail if prices experience drastic jumps Distribution of Option Payoffs Unlike linear derivatives (futures, forwards) payoffs on options are intrinsically asymmetric Not necessarily because of distribution of the underlying factor, which is often symmetric, but rather is due to the exposure profile. Long positions in options have a positive gamma positive skewness or long right tails Short positions in options are short gamma hence have negative skewness or long left tails VAR formulas for simple option position Assuming normal distribution VAR of underlying asset is VAR(dS) = S(dS/S)
corresponds to desired confidence level

Seite 54 von 64

Linear VAR of option is VAR1(dc) = x VAR(dS) Quadratic Var of option is VAR2(dc) = x VAR(dS) 1/2 x VAR(dS) Long option positions have positive gammas and hence lower risk than using a linear model Negative gammas translate into higher VARs For such options the extent of nonlinearity matters Nonlinearity also depends on the horizon For plain-vanilla options, the linear approximation may be adequate as long as the VAR horizon is kept short. For more exotic options, or longer VAR horizons, risk managers must account for nonlinearities

15 VAR Methods
VAR: Local versus Full Valuation Local valuation methods make use of the valuation of the instrument at the current point, along with first and perhaps second partial derivatives Full valuation methods reprice the instruments over a broad range of values for the risk factors Mapping is first step and consists of replacing each instrument by its exposures on selected risk factors Local valuation models also are known as analytical methods Riskmeasurement o Local valuation Linear Models Full covariance matrix Factor models Diagonal models Nonlinear Models Gamma convexity o Full valuation Historical simulation Monte Carlo simulation Local Valuation VAR was born from recognition that we need an estimate that accounts for various sources of risk and expresses loss in terms of probability. Extending the duration equation to the worst change in yield at some confidence level dy we have (Worst dP) = (-D*P) * (Worst dy)
D* modified Duration

Seite 55 von 64

For a long position in the bond , the worst movement in yield is an increase at say, the 95% confidence level -> this will lead to fall in the bond value at the same confidence level Approach is called local valuation because it uses information about the initial price and exposure at the initial point VAR for bond is VAR(dP) = | -D*P | * VAR(dy) Advantage of approach is its simplicity -> distribution in price is the same as that of the change in yield -> this is particularly convenient for portfolios with numerous sources of risk , because linear combinations of normal distributions are normally distributed Full Valuation To take into account nonlinear relationships, one would have to reprice the bond under different scenarios for the yield (Worst dP) = P[y0 + (Worst dy)] P[y0] Approach is called full valuation, because it requires repricing of asset It is more precise and more complex as simple linear valuation Delta-Gamma Method Would like to keep simplicity of local valuation while accounting for nonlinearities in payoff patterns Using Taylor expansion where the second order term involves convexity C dP ~ aP/aydy + ()aP/ay(dy) ) = (-D*P)dy + (1/2) CP(dy) valuation is still local, because we only value bond price once . the first and second derivatives are also valued at the local point Because price is a monotonic function of underlying yield, we can use Taylor expansion to find the worst down move in the bond price from the worst move in the yield = dy* = VAR(dy) (Worst dP) = P(y0 +dy*) P(y0) ~ (-D*P)(dy*) + (1/2)(CP)(dy*) Leads to simple adjustment for VAR VAR(dP) = |-D*P| * VAR(dy) (1/2)(CP) * VAR(dy) More generally, this method can be applied to derivatives, for which we write the Taylor approximation as df = dS +( )dS is the second derivative, or Gamma, like convexity For long call option, the worst value is achieved as the underlying price moves down by VAR(dS) With > 0, > 0 VAR for this derivative is now VAR(df) = || *VAR(dS) (1/2) * VAR(dS) This method is called delta-gamma, because it provides analytical, second-order, correction to the delta-normal VAR. This simple adjustment only works for monotonic (1:1 relation) payoff functions More generally the delta-gamma-delta VAR method involves, Seite 56 von 64

 First: computing the moments of df Second: choosing the normal distribution that provides the best fit for these moments Improvement brought about by this method depends on the size of second-order coefficient, as well as the size of the worst move in the risk factor For Forward contracts = 0 no point in adding second-order terms Similarly for most fixed-income instruments over a short horizon the convexity effect is relatively small -> ignore VAR Methods: Overview Mapping Portfolio consist of large number of instruments very complex to model each instrument separately first step is mapping = replacing the instruments by positions on a limited number of risk factors the distribution of the portfolio return is then derived from the exposures and movements in these risk factors Delta-Normal Method Is simplest VAR approach, it assumes that the portfolio exposures are linear and the risk factors are jointly normally distributed is a local valuation method Because portfolio return is a linear combination of normal variables => is normally distributed Using matrix notations, the portfolio variance is given by (Rp,t+1) = xtt+1xt t+1 is the forecast of the covariance matrix over the horizon If portfolio volatility is measured in dollars, VAR is directly obtained from the standard normal deviate that corresponds to confidence level c VAR = (Rp,t+1) Is also called diversified VAR, because it accounts for diversification effects Undiversified VAR is simply the sum of individual VARs for each risk factor. It assumes that all prices moves in the worst direction simultaneously = unrealistic RiskMetric approach is similar to delta-normal approach. Only difference is that risk factors returns are measured as logarithms of the price ratios, instead the rate of returns Main benefit = appealing simplicity = drawback, because delta-normal method cannot account for nonlinear effects; it may also underestimate the occurrence of large observations Historical Simulation Method Is a full valuation method, it consist of going back in time and applying current weights to a time-series of historical asset returns -> replays tape of history with current weights Define the current time as t; we observe data from 1 to t. The current portfolio value is P t , which is a function of current risk factors Pt = P [f1,t, f2,t, ... fN,t] We sample the factor movements from historical distribution, without replacement fik = { fi,1, fi,2, .,fi,t} From this we can construct hypothetical factor values, starting from the current one fik = fi,t + fik Which are used to construct a hypothetical value of current portfolio under the new scenario Seite 57 von 64

Pk = P[f1k, f2k, , fNk] We now compute changes in portfolio values from the current position Rk = (Pk Pt)/Pt We sort the t returns and pick the one that corresponds to the cth quintile, Rp(c) VAR is obtained from the difference beween the average and the quantile VAR = AVE[Rp] Rp(c) Advantage of this method is that it makes no specific distributional assumption about return distribution, other than relying on historical data. Is advantage about normal distribution, because historical data typically contains fat tails Main drawback is its reliance on a short historical moving window to infer movements in market prices. If window does not contain likely market moves that are likely, it may miss some risks Monte Carlo Simulation Method Is similar to historical simulation, except that the movements in risk factors are generated by drawings from some prespecified distribution fk ~ g(), k = 1, . K Where g is the joint distribution and the required parameters The risk manager samples pseudo-random numbers from this distribution and then generates pseudo-dollar returns as before. Finally the returns are sorted to produce the desired VAR This method is most flexible but carries enormous computational burden It requires users to make assumptions about the stochastic process and to understand the sensitivity of the results of these assumptions. Thus it is subject to model risk MC creates inherent sampling variability, because of randomnation -> different random numbers will lead to different results ->may take large number of iterations to converge to stable VAR measure. When all risk factors have normal distribution and exposures are linear the converge VAR should be the same as from Delta-normal VAR Comperison of Methods Features Valuation Distribution Shape Extreme events Implementation Ease of computation Communicability VAR precision Major pitfalls Delta-normal Linear Normal Low probability Yes Easy Excellent Nonlinearities, tails Historical simulation Full Actual In recent data Intermediate Easy Poor with short window fat Time variation in risk, unusual events Monte simulation Full General Possible No Difficult Good with iterations Model risk Carlo

many

Limitations of VAR System Goal of risk measurement systems is to describe the distribution of potential losses on portfolio. VAR is single summary measure of dispersion in portfolio returns and consequently has limitations that should be obvious Cannot be viewed as a worst-loss measure -> instead should be viewed as a measure of dispersion that should be exceeded with some regularity confidence level

Seite 58 von 64

Does not describe the extent of losses in the left tail -> Instruments, such as short positions in options could generate infrequent but extreme losses -> to detect such vulnerabilities distribution of losses beyond VAR should be examined as well Traditional application of VAR models, such as historical simulation, involves moving windows(typically one to three years of data) -> window may not represent range of potential movements in risk factors -> stress tests are needed Implementation of VAR often requires simplifications, obtained by mapping positions on the selected risk factor -> risk managers should be cognizant of weaknesses in their risk system During credit crisis that started 2007, risk management systems failed at many banks Some banks suffered losses that were much more frequent and much worse than anticipated UBS suffered 29 exceptions (worse than VAR) instead of experience of 2 to 3 exceptions Due to number of factors 1. Markets were extremely volatile after extended period of relative stability Firms entered markets 2007 with high leverage-> when volatility started to rise of exceptions started too In addition to these effect banks experienced large losses on super senior triple A rated tranches of securities backed by subprime mortgages Investing in super senior tranches can be viewed as selling out-of-the-money put options As long as the real estate market went up -> default rates on subprime debt was low and super senior debt was safe -> but as real estate market corrected sharply -> put options moved intothe money -> led to large losses Differences in Risk Management Practices Practice Winners Busines Model Avoided CDOs, SIVs Organizational structure Cooperative Firm-wide risk analysis Shared information across the firm Valuations Developed in-house expertise Management of liquidity Charged business lines for liquidity risk Risk measurement Qualitative and quantative analysis Varied assumptions Tested correlations

Losers Exposed to CDOs SIVs Hierarchical No prompt discussion of risks across the firm Relied on credit ratings Did not consider contigent exposures Strict model application Mapped to corporate AAA No test of correlations

22 Credit Derivatives and Structured Products

Credit derivatives are latest tool in management of portfolio credit risk. Credit derivatives are contracts whose value derives from the credit risk of an underlying obligor, corporate, sovereign or multiname. They allow exchange of credit risk from one counterparty to another. Initially grew from need of banks to modify their credit risk exposure Like other derivatives they can be traded standalone or embedded in some other instrument such as the credit-linked note. This market has led to the expansion of structured credit products, through which portfolios of credit exposures are repackaged to better suit the needs of investors Introduction Market size From 1996 to 2007 market for credit derivatives is estimated to have grown from $40 billion to 62000 billion, all of which is currently traded in OTC markets

Seite 59 von 64

Gross exposures are very high because dealers had a practice of not cancelling existing trades but instead simply added new ones with offsetting characteristic. -> this practice created backlog of paperwork and increases operational risk as well as counterparty risk As a result , the industry is starting to implement portfolio compression = prozess that reduces the overall size and number of items in credit portfolios without changing the risk parameters of the portfolio Markets for Exchange of Risks Credit derivatives have grown quickly because they provide an efficient mechanism to exchange credit-risk. While modern banking is build on sensible notion that a portfolio of loans is less risky than single ones, banks still tend to be too concentrated in geographical or industrial sectors, because of comparative advantage in relationship banking Difficult to lay off risk exposure Bond insurance is a contract between a bond issuer and a guarantor to provide additional payment should issuer fails to make full and timely payment A letter of credit is a guarantee by a bank to provide a payment to a third party, should the primary credit fail on its obligations Call feature in corporate bonds involves in an option on the risk-free interest rate as well as the credit-spread -> borrower may call back bond should credit rating improve Long position in corporate bond = to long position in risk-free bond plus a short position in a credit default swap New is transparency and trading made possible by credit derivatives Corporate bonds are difficult to short Types of Credit Derivates Are over the counter contracts that allow credit risk to be exchanged across counterparties Classification after: The underlying credit, which can be either single entity or a group of entities The exercise conditions, which can be a credit event The payoff function, which can be a fixed amount or a variable amount with a linear or nonlinear payoff Market includes plain-vanilla credit default swaps, total return swaps, credit spread forwards and options. These instruments are bilateral OTC contracts Credit Default Swaps Definition In a credit default swap contract, a protection buyer pays premium to protection seller, in exchange for payment if a credit event occurs. Premium payment can be a lump sum or periodic. The contingent payment is triggered by credit event on the underlying credit = purest form of credit derivative, as settlement payment occurs only in default case Contracts are really options and not swaps! Difference is that cost of option is paid in instalments instead of up front, if payment is paid up-front than contracts are called default put options, annual payment is also known as CDS spread. CDS are quoted in terms of the annual spread, with payment made on quarterly basis Settlement Credit events must be subject to precise definition -> no interpretation if event is credit event or not -> otherwise transactions can create legal risk Payment reflects the loss to the holders of the reference asset when credit default event occurs. Seite 60 von 64

Can take number of forms Cash settlement = payment equal to strike minus the prevailing market value of underlying bond Physical delivery = of defaulted obligation in exchange for fixed payment Lump sum = fixed amount based on some pre-agreed recovery rate = > payoff on CDS is Payment = Notional * Value Indicatorfunction(Creditevent) of this payment per unit of notional *

Swap spread reflects both the probability of default and the loss given default Slight variant is the binary credit default swap, which pays fixed amount, if credit event occurs. These two contracts can be combined to extract a market-implied estimate of the recovery rate With physical delivery: List of bonds that can be delivered is defined -> usually protection buyer chooses cheapest bond which creates a delivery option Cash settlement can be conducted through an auction, which defines the recovery rate Pricing CDS contracts can be priced by considering the present value of the cash flows on each side of the contract Fair value of the spread is the number that set the initial value of CDS contract to zero V =(PV Payoff) s(PV Spread) = (kt(1-f)PVt) s(St-1PVt) Abstracting from counterparty risk the CDS spread should be approximately equal to the difference between the yield on a corporate bond issued by same obligor and risk-free yield of same maturity. Otherwise arbitrage is possible Basis between the CDS spread and cash yield is slightly positive -> influenced by supply and demand Counterparty Risk Entering CDS does not eliminate credit risk entirely; instead the protection buyer decreases exposure to the reference credit but assumes new credit exposure to the CDS seller Protection will be effective with a low correlation between the default risk of underlying credit and of the counterparty. Contracts may involve the posting of collateral from protection seller. Like options these instruments are unfunded= each party is responsible for making payments without recourse to other assets. Funded instruments = payment could be used to settle any potential credit event Other Contracts CDS Variants CDS can be written on multiple names, e.g. first-of-basket-to-default swap gives protection buyer to deliver one and only one defaulted security out of a basket of selected securities-> protection buyer has more choices -> more expensive Seite 61 von 64

Price also depends on correlation between default events -> low correlation = more expensive, high correlation = small swap rate With an Nth-to-default swap, payment is triggered after n defaults in underlying portfolio CDS indices are widely used to track performance of this market Total Return Swaps = contract where protection buyer makes a series of payments linked to the total return of reference asset, in exchange the protection seller makes a series of payments tied to a reference rat, such as the yield on an equal Treasury issue plus a spread If price of asset get down protection payer receives payment from counterparty, if it goes up payment is due in other direction Swap is tied to changes in market value of the underlying asset and provides protection against credit risk in a market-to-market framework. For buyer it removes all economic risk of underlying asset without selling it. TRS involves credit and market risk, the latter reflecting pure interest rate risk Credit Spread Forward and Options Instruments are derivatives whose value is tied to underlying credit spread between risky and risk-free bond In a credit spread forward contract buyer receives difference between the credit spread at maturity and an aged-upon spread if positive Payment = (S-F) * MD * Notional Payment =[P(y+ F,t) P(y + S,t)] * Notional In a credit spread option contract, the buyer pays a premium in exchange for the right to Put any increase in the spread to the option seller at a predefined maturity Payment = Max(S-K,0)*MD*Notional Structured Products Creating Structured Products Can be defined as instruments created to meet specific needs of investors or borrowers that cannot met with financial instruments Typical example is retail demand for investments that participate in the appreciation of stock markets but also preserve capital. Payoff profile of the product can be replicated from a combination of existing, or sometimes new instruments. In this case by investment in a risk-free bond with notional equal to guaranteed capital plus long positions in a call option. This instrument is a principal-protected note and can be indexed to variety of markets Credit-Linked Notes = Are structured securities, which combine a credit derivate with a regular bond. Buyer of protection transfers credit risk to an investor via intermediary bond-issuing entity This entity could be the buyer itself or a special purpose vehicle Provider CL Note Investor

Asset

Collateralized Debt obligation Seite 62 von 64

Collateralized mortgage obligations brought mortgage-backed securities to the masses by repacking their cash flows into tranches with different characteristics Collateralized debt obligations are securities backed by pool of debt Collateralized bond obligations and collateralized loan obligations are backed by bonds and loans First step is to place a package of corporate bonds in a special purpose vehicle->multiple tranches issued by SPV with a specified waterfall structure or priority of payments to the various tranches (senior, mezzanine = intermediate, subordinated/equity) In the simple structure: SPV = passive entity-> redistributes cash flows according to well defined rules (no other management action) Attachment point and detachment point, e.g. tranche C absorbs losses from 3 to 10% Bottom =equity tranche = not rated -> return can be very high (if no default) due to leverage>exposed to the first dollar loss in the portfolio-> special conventions apply to trading: Investor pays notional, protection seller receives spread = running spread and an up-front fee (quoted in %, typically 40% for investment grade CDOs) -> reduction of notional due to cumulative losses = running spread applied to new notional Internal models by rating agencies to rate senior tranches based on the probability of shortfalls due to default Basic loss of conservation for resulting package: For the underlying and resulting securities same cash flows for each point time (part of transaction cost) Same total market value and same risk profile, both for interest rate and default risk Weighted duration of final package = debt of underlying securities Expected default rate(averaged by market values) must be the same -> some tranches less risky others bear more risk->structured so that risk of most of the tranches < risk of collateral -> remaining residual tranche is more risky(retained by institution sponsoring CDO) CDO Market Balance sheet and arbitrage CDOs Classified by purpose: Balance sheet CDOs: primarily goal= move loans off the balance sheets, off commercial banks , to lower regulatory capital requirements Arbitrage CDOs: to capture the spread between portfolio of underlying securities and that of highly rated, overlying, tranches->CDO senior tranches relatively save due to diversification effects, pay tight spread over LIBOR, arbitrage profit goes into the equity tranche(+management and investment banking fees) Cash flow and synthetic CDOs Credit risk transfer can be achieved by cash flow or synthetic structure Traditional, or funded, cash flow CDOs: physical assets are sold to SPV, underlying cash flows are used to back payments to issued notes Synthetic CDOs: credit risk exposure achieved with credit default swaps->create higher yields by first, funding or placing the initial investment in default-free / Treasury securities, and second, selling group of CDSs to replicate a cash flow CDO Advantages of synthetically CDOs Easier to manage as cash flow CDOs -> bankruptcy of one underlying credits = management of cash flow SPVs takes part in bankruptcy process->short CDS position cash settled: no need for involvement Seite 63 von 64

 Issue does not need to be fully funded o Full capital structure CDO: total notional amount of notes issued = total notional amount of the underlying portfolio-> fully funded o Single tranche CDO: bank and investor agree on terms of deal(size, credit rating, underlying credits) -> bank holds rest of capital structure (no placement) Cash flow and market value CDOs Cash flow CDOs: payments to investors solely from collateral cash flows Market value CDOs: payments from collateral cash flows as well as sales of collateral ->market value of collateral < specified level = payments to equity tranche suspended -> more flexibility for portfolio manager Overcollateralizationratios (OC)= used to analyse quality of credit structures -> measures how many times the collateral can cover the SPV liabilities OC = V/D
V market value of assets(= market value CDOs/PV of Total assets(cashflow CDOs) D par value of liabilities

Must be high enough to ensure sufficient coverage of liabilities Alternatively: par value of cumulative tranches (start =TOP) must be capped below the market value of assets times an advanced rate structure failing the overcollateralization test risks downgrading -> cured by selling some of assets and repaying some of the tranches or issuing more equity Interest rate coverage ratio(IC) Analyse quality of credit structures-> computed as the total interest payment to be received by collateral divided by interest liability of each tranche and above Static and managed CDOs Difference in management of asset pool: static CDOs: asset pool is basically fixed managed CDOs: portfolio manager is allowed to trade the underlying assets -> usual benefits and (disadvantage) of active management: ability to unwind assets with decreasing credit quality, to buy undervalued securities, and to sell overvalued securities -> investors face credit risk + management risk ( + management fees) Other Products CDO-squared structure: CDO invest in CDO tranches instead of individual credits -> main benefit = greater degree of diversity ABCDs: market trades credit default swaps on Asset Backed Security (ABS) tranches -> assets mostly backed by home equity and commercial property loans -> long on ABCDs = equivalent to acquiring protection or shorting the security Constant proportional debt obligation (CPDO); offer protection on portfolio of corporate credits -> transaction is highly levered and dynamically adjusted, getting rid of credits that deteriorate over time and changing the leverage as spreads vary -> look attractive in an environment of stable or falling credit spreads, but riskprofile is difficult to assess due to dynamic nature Seite 64 von 64