Page 1 of 30

Modeling the Volatility Smile

Emanuel Derman Columbia University

October 27, 2006

Stanford.Smile.fm

October 21, 2006

Page 2 of 30

The Implied Volatility Smile/Surface

Black-Scholes

S&P 1995

B way & 116th, 2004

Black-Scholes implied volatilities for equity indices: ( S, t, K, T ) Term structure of strike and expiration, which change with time and market level Always a negative slope w.r.t strike for equity index options What model replaces Black-Scholes?

Stanford.Smile.fm

October 21, 2006

Page 3 of 30

The Smile is a Post-Crash Phenomenon

Stanford.Smile.fm

October 21, 2006

Page 4 of 30

Current Equity Index Smile

as a function of K/S

ln ( K S ) as a function of moneyness -------------------- or

it steepens with maturity
Foresi & Wu: Journal of Derivatives, Winter 2005

Stanford.Smile.fm

October 21, 2006

Page 5 of 30

Negative Correlation Between Level and Slope

Correlation decreases with expiration

Cross Correlations between Volatility Level and Smirk Slope

0 0.1 0.2 0.3

TO

RE

EXHIBIT 5

PR

0.4 0.2

GA

0.4 0.5 0.6 0.7 0.8 0.9

Corr( c0, c1)

0
0.2 0.4 0.6 0.8

Corr(c0,c1)

IT

IS

IL

LE

0.5

1.5

2.5

3.5

4.5

0.5

1.5

2.5

3.5

4.5

Maturity in Years

Maturity in Years

Lines denote the cross-correlation estimates between the volatility level proxy (c0) and the volatility smirk slope proxy (c1). The left panel measures the correlation based on daily estimates, the right panel measures the correlation based on daily changes of the estimates.

Stanford.Smile.fm

October 21, 2006

Page 6 of 30

Volatility of Implied Volatility Decreases with Expiration

Short-term implied volatilities are more volatile. This suggests mean reversion or stationarity for the instantaneous volatility of the index.

Stanford.Smile.fm

October 21, 2006

Page 7 of 30

Other Markets
VOD Imp Vol (12/27/01)
70% 60% Imp Vol 50% 40% 30% 0.5 0.75 1 Strike/Spot 1.25 1.5 1 Mon 3 Mon 6 Mon 12 Mon

single stock implied vols

USD/EUR

MXN/USD

JPY/USD

currency implied vols

ATM Strike = 0.90 9.85 123.67

Stanford.Smile.fm

October 21, 2006

Page 8 of 30

The Smile is a Problem for Hedging

Black-Scholes is being used as a quoting mechanism for vanilla options (cf. yield to maturity for quoting bond prices) but is not the correct valuation/hedging model. If the valuation formula is wrong, then the hedge ratio is wrong, therefore the price doesnt correspond to the correct replication formula. Whats the right way to hedge index options? Estimate of hedging error for an S&P index option with S ~ 1000 and T = 1 year

dC BS ( S, t, K, T, ) C BS C + = --------------------------------------------- = S dS S C S T ---------- 400 2 0.02 --------- 0.0002 S K 100

Eq.1.1

C 400 0.0002 = 0.08 S

For a typical move S 10 S&P points the hedging error = 0.8 points

1 S 1 S The P&L earned from perfect hedging is -------- -------------- -------- -------- ( 50 ) = 0.25 points , 2 2 200 S T
which is dwarfed by the error.

Stanford.Smile.fm

October 21, 2006

Page 9 of 30

The Smile is a Problem for Valuing Exotics

Since Black-Scholes is inappropriate, we dont know how to value all sorts of exotic averages: barriers, averages, lookbacks which effectively involve multiple strikes and therefore multiple implied volatilities. Which one do you use in a Black-Scholes world? How can we fix/replace/alter/correct Black-Scholes?

Stanford.Smile.fm

October 21, 2006

Page 10 of 30

What Could Cause the Smile?

Behavioral Causes Crash protection/ Fear of crashes) [Katrina] Out-o- money puts rise relative to at-the-money; so do realized volatilities Expectation of changes in volatility over time Support/resistance levels at various strikes Dealers books tend to be filled by the demand for zero-cost collars

Structural Causes: Violations of Black-Scholes assumptions Inability to hedge continuously Transactions costs Local volatility ( S, t ) ; leverage effect; CEV models Stochastic volatility Jumps/crashes

Stanford.Smile.fm

October 21, 2006

Page 11 of 30

The Principles of Modeling

Physics is about the future; finance is mostly about expectation of the future. Absolute valuation Newton in physics; no such thing in finance Relative valuation: The only reliable law of finance If you want to know the value of something (illiquid), find the market price of something similar and liquid. Similar: same payoffs under all (?) circumstances. static replication is best needs no systems, virtually no assumptions. dynamic replication is next best but needs uncertain replication models, calibration, liquidity assumptions and a big investment in technology. If you cant do either of those, then markets are incomplete and you need a utility function.

A model is only a model: capture the important features, not all of them.

Stanford.Smile.fm

October 21, 2006

Page 12 of 30

Static Replication of Exotic Options

The best way to handle valuation in the presence of a smile is to use no model at all. For terminal payoffs:

Any of these payoffs can be exactly duplicated by a linear combination of forwards, bonds, puts and calls with a range of strikes, and hence exactly replicated at a cost known if we know the volatility smile. Replication will hold irrespective of jumps, volatility, etc. -- only problem is counterparty risk. Try to do approximately the same for other exotic options.

Stanford.Smile.fm

October 21, 2006

Page 13 of 30

Dynamic Replication: Local Volatility Models

dS ----- = ( S, t )dt + ( S, t )dZ S
risk-neutral pricing.
Eq.1.2

( S, t ) is a deterministic function of a stochastic variable S. Can still replicate options, still do

Calibration: ( S, t ) ( S, t, K, T ) ? Even if there is a solution, does the local volatility describe the asset? Why would local volatility make sense: Leverage effect: as assets approach liabilities, equity volatility increases Fear: increase in implied volatility as equity declines CEV models: dS = ( S, t )dt + S dZ but they are parametric and cannot match the implied volatility surface

Stanford.Smile.fm

October 21, 2006

Page 14 of 30

The Smile in Local Volatility Models: First Look

Interest rates: long rates are averages of expected future short rates between today and maturity In local volatility models implied volatilities are averages of expected future volatilities between spot and strike, between today and expiration.

( S, K ) = average vol between S and K

strike K local volatility (S,t) spot S
T

(S) + (K) ( S, K ) -----------------------------2

S=K

1 -- ( S ) 2 S=K

Local volatilities predict \ the change of implied volatility with stock price, and that local volatility varies twice as fast with spot as implied volatility varies with strike.

Stanford.Smile.fm

October 21, 2006

Page 15 of 30

Local Volatility Models

(K + S)
Lets draw this for a roughly linear skew in the following figure.
( 80, 80 )

( S, K )
( 80, 100 ) ( 100, 80 )

( 100, 100 )

S = 80 S = 100 K

80

100

Smile moves up when stock price moves down.

Stanford.Smile.fm

October 21, 2006

Page 16 of 30

The Relation Between Local and Implied Vols

Think of implied volatilities as a complex probe/X-ray of the interior of the local volatility surface. Can one deduce local volatilities from implied?

This is an inverse scattering problem.

Stanford.Smile.fm

October 21, 2006

Page 17 of 30

Inverse Scattering: Schematic Calibration

Derman-Kani binomial method

Stanford.Smile.fm

October 21, 2006

Page 18 of 30

Dupire Continuous-Time Equation

Recall the Breeden-Litzenberger formula: risk-neutral probability/Arrow-Debreu price is a butterfly spread:

p ( S, t, K, T ) = e

r(T t)

2 2

(C ( S, t, K, T ))
K

We can go further than this, and actually find the local volatility ( S, t ) from market prices of options. For zero interest rates and dividends, the local volatility at S T = K can be determined from the derivatives of option prices:

C ( S, t, K, T ) T ( K, T ) --------------------- = -----------------------------------2 2 2 C K 2 K
2

calendar spread ~ variance x prob butterfly spread ~ variance

Stanford.Smile.fm

October 21, 2006

Page 19 of 30

Example: Local Volatility Extraction

Local volatilities are the analogs of forward/sideways rates

local vol vs. spot S

implied vol vs. strike

Implied and Local Vols at 1 years 0.25 0.2 0.15 0.1 0.05

local vol vs. spot S implied vol vs. strike K

spot or strike
60 80 100 120 140

Stanford.Smile.fm

October 21, 2006

Page 20 of 30

Local Volatility

In practice, we dont have a continuous implied volatility surface so we cannot use these relations blindly. Needs more sophisticated methods. Nice approximate harmonic average relation for short expirations due to Roger Lee:

x = ln ( S K ) 1 1 1 ---------- = -- ---------- dy x (y) (x)

o x

Stanford.Smile.fm

October 21, 2006

Page 21 of 30

Stochastic Volatility Models

dS = S ( S, V, t )dt + S ( S, V, t )dZ t dV = V ( S, V, t )dt + V ( S, V, t )dW t V = E [ dWdZ ] = dt
2

Can replicate only if you can trade volatility/options as well as stock. Option prices are risk-neutral, volatility is not -- need market price of risk for volatility. You must hedge with stock and options - difficult! Can one really know the stochastic PDE for volatility? A tall order. (Rebonato). Correlation is even more stochastic than volatility.

Stanford.Smile.fm

October 21, 2006

Page 22 of 30

The Smile in Stochastic Volatility Models

Poor mans point of view: Black-Scholes with the volatility in one of two regimes: Take expectation of prices:

1 C SV = -- [ C BS ( S, K, H ) + C BS ( S, K, L ) ] 2

H L

Because the Black-Scholes equation is homogeneous in S and K, we can write

K C SV = Sf --- SC BS ( S, K, ) S
So

K = g --- S

at the money S K

Just the inverse of the relation in local volatility models.

Stanford.Smile.fm

October 21, 2006

Page 23 of 30

The Smile in Stochastic Volatility Models

(S K)
Smile moves up when stock moves up, if volatility doesnt change.

( S, K )
( 100, 80 ) ( 100, 100 ) ( 80, 80 )

( 100, 120 ) ( 80, 100 )

S = 100 S = 80 120 K

80

100

Stanford.Smile.fm

October 21, 2006

Page 24 of 30

Stochastic Volatility PDE

Volatility is mean-reverting: d
2

= ( m )dt + dW r
Su,u Su,d

Heston, etc.

S,

Sd,u Sd,d

V V V 1 V 2 2 1 V 2 V + -qS + rS + ( S, , t ) rV = 0 S + -q + 2 2 S S t 2 S2
Risk neutrality: volatility drift or market price of risk is determined by options prices being riskneutral.
Stanford.Smile.fm October 21, 2006

Page 25 of 30

Stochastic Volatility Characteristic Solution

1 2 T = -- t dt T
0 2 T

average variance along a path

V =

p ( T ) B S ( S, K, r, T, T )
T

the Mixing Formula for = 0

H L

Stanford.Smile.fm

October 21, 2006

Page 26 of 30

Stochastic Volatility Smiles:

symmetric for = 0 , because of fat tails; highly curved for short expirations; flat for long expiration as mean reversion wins out. When 0 smile is asymmetric. You need very high volatility of volatility to get a steep short-term smile. If you cannot hedge with options, then the optimal pure stock hedge ratio looks a lot like the local volatility hedge ratio, i.e. < BS when < 0

Imp Vol = -1 > -1

=0

Local volatility is the average of all stochastic volatilities at a definite future stock price and time.

strike

Stanford.Smile.fm

October 21, 2006

Page 27 of 30

Jump-Diffusion Models
In fact, not only is volatility stochastic, but more seriously, stock prices move discontinuously. Geometric Brownian motion isnt the right description. Merton: Poisson distribution of unhedgeable but diversifiable jumps.

S+J

S 1- SK

BS

BS

C JD

2 J + 1 2 2 -n J + 1 n - 2 J 2 n J ( ) e 2 J 1 = e -------------CBS S, K, , + ---------, r + --------------------------- n! n=0

Stanford.Smile.fm

October 21, 2006

Page 28 of 30

Jump-Diffusion Smiles

Steep realistic short-term smiles from the instantaneous jump.
0.29

Implied volatility under jump diffusion

Long-term smiles tends to be flat as diffusion

implied volatility

0.285 0.28 0.275 0.27 0.265 0.26 0.255 0.25 0.245 0.24 0.235 "0.1" "0.4" "100"

overwhelms the jumps.

A higher jump frequency helps produce a steeper smile at expiration, because the jumps tends to happen more frequently and therefore are more likely to occur in the future as well.

-0.1 -0.1 0 0.05 0.1 Stochastic volatility, in contrast, has difficulty ln(K/S) producing a steep short-term smile because the volatility diffuses, and therefore doesnt change too much initially. The smiles in a stochastic volatility model are more pronounced at longer expirations.

Stanford.Smile.fm

October 21, 2006

Page 29 of 30

Pragmatic Interpolation Methods

Practitioners think of options models as interpolating formulas that take you from known prices of liquid securities to the unknown values of illiquid securities: options; convertible bonds.

An exotic option is a mixture of ordinary options. Find a static hedge composed of a portfolio of vanilla options that approximately replicates the payoff or the vega of the exotic options. Then figure out the value of the portfolio of vanilla options in a skewed world.

Stanford.Smile.fm

October 21, 2006

Page 30 of 30

Summary

There is no right model. The best you can do is pick a model that mimics the most important behavior of the underlyer in your market. Then add perturbations if necessary. Examine the value of an option in a variety of plausible models. Try to be as model independent as possible. There is lots of work to be done on modeling realistic distributions and the options prices they imply.

Stanford.Smile.fm

October 21, 2006