Is there an “Historical Skew” ?

January 2004

Nicolas Blanc
+33 (1) 40 14 76 75
nicolas.blanc@bnpparibas.com
Assessing The Skew With Historical Returns

„ Is the current skew fair compared to the past stock

returns behavior?

„ Which strike or maturity provides the best values?

„ Which volatility curve arbitrage is the most

promising?

„ What is the fair implied volatility of a CB with no listed

options on the underlying?

„ What is the fair implied volatility for an option on a

basket of assets?

Page 2
Executive Summary

„ Skew definition

„ Skew and Risk Neutral Density

„ General pricing models

„ “Fair skew” models

„ Risk Neutral Historical Density

„ Historical Skew

Page 3
The Black & Scholes model

„ Asset follows a log-normal process

dSt
= µdt + σdWt
St

„ Theoretical option prices and hedge ratios:

C(K ,T ,σ ) = S0 N(d1 ) − Ke − rT N(d 2 ) ∆(K ,T ,σ ) = N(d 2 )

„ If the option market obeys the B&S model, there exists

σ such that

Market Price [Call(K,T )] = C (K ,T ,σ )

„ There is a direct estimator for the volatility parameter:

The historical volatility

Historical vol
Historical Returns Fair option price
B&S Model

Page 4
The Skew Is A Correction To B&S

„ In reality, option market prices do not result from a

unique B&S volatility
„ Market Price [Call(K1,T1 )] = C (K1,T1,σ 1 )
„ Market Price [Call(K 2 ,T2 )] = C (K 2 ,T2 ,σ 2 )
„ σ1 ≠ σ 2

„ Although the model is invalidated, option traders use

the same pricing function, with a volatility correction
„ Market Price [Call(K,T )] = C (K ,T ,σ (K ,T ))
„ K → σ (K ,T ) is called the skew
„ T → σ (K ,T ) is called the term structure

„ Historical volatility ?

Page 5
Typical Skew Pattern

„ Out of the money puts appear more expensive than

ATM puts

35,00

30,00

25,00

20,00

15,00

10,00
3000 3500 4000 4500 5000 5500 6000 6500 7000 7500

Page 6
Pricing An Equity Derivative

„ General case
General Pricing
“Skewed” model Price any (exotic)
Option Prices + Calibration derivative
procedure

„ Specific case: European products

“Skewed” Risk Neutral

Density Price a European
Option Prices contingent claim
Estimation

„ Relationship between historical asset returns and the

current skew is little explored

Historical RNHD Fair

Asset returns Historical skew Skew
Page 7
Skew & Risk Neutral Density

„ From the skew we may price any European derivative

S→F(S) at maturity T

Butterfly Spread around strike K

1
Short 2/dK Call Strike K
∂ 2C (K ,T )
Long 1/dK Call Strike K + dK ≈ dK
∂K 2
Long 1/dK Call Strike K - dK

K-dK K K+dK

F(Ki) ∑ F(K )BS (K

i =1
i i −1, K i , K i +1 )

∂ 2C (K ,T )
∞
 → ∫ F (K )
dK →0
dK
−∞
∂K 2

K1 K2 Kn
Page 8
Risk Neutral Density

„ The skew defines a probability density

Fair pricing ⇔ discounting the average option payoff
under the risk neutral distribution

1,2
∂ 2C (K ,T )
ϕT (K ) = e
rT

∂K 2
1
Option with payoff F(S) at T,

0,8
Arbitrage Price
= e -rT ∫ F(s )ϕ (s )ds
0,6 = e -rT EQ [F(S )]
where Q is given by ϕ
0,4

0,2

0
0 10 20 30 40 50 60 70
Underlying Stock
Page 9
Interpretation: Risk Neutral Density

„ Compared to the B&S model, the skew introduces

asymmetry in the future asset returns RN-distribution

0,0018

0,0016
Normal density function
0,0014

0,0012

0,001

0,0008

0,0006
Skew-implied
density function 0,0004

0,0002

0
-50% -40% -30% -20% -10% 0% 10% 20% 30% 40% 50%

Log Returns
Page 10
Next Step: Skew-Compatible Asset Process

„ It is possible to construct processes whose density at

each date t is given by the option-implied ϕt

dS
= (r − d )t + σ (S, t )dWt
S
„ Case of local volatility models ∂C ∂C
2 + (r − d )K + dC
σ (S, t ) = ∂T ∂K
2 ∂ C
2
„ Beyond: K
∂K 2
„ Jump models
„ Stochastic volatility models

„ None of these approaches provide an easy connection

between historical asset returns and the current skew

Page 11
Risk Neutral Historical Density

„ Derman / Zou (99)

„ Arbitrage pricing: Skew ⇒ “Non Gaussian” RND

„ Historical RND ⇒ Skew

„ Estimation
„ Smoothing
„ Risk Neutralization

Page 12
Risk Neutral Historical Density

„ Empirical distribution of Rt = Log(St / St-T)

2800
T
2600

2400

2200

2000

1800
01/03 07/03

Discretization of Log returns Page 13

Risk Neutral Historical Density

„ Kernel smoothing
T
 x − Rt 
(R1,L, RT ) → f (x ) = 1 ∑ K  
υ.T t =1 υ 
8

Discrete density

4
f(x): Continuous density

Log Variation
0
-60% -40% -20% 0% 20% 40% 60%

Page 14
Risk Neutralization

„ Historical returns have a trend

∞
F (S ) = S ⇒ S0 = e − rT
∫ sϕ (s )dx
−∞

3500

3100

2700

2300

If we stick to simple HD, we

would have different forward
1900
prices

1500
12/03/03 12/06/03 12/09/03 12/12/03

Page 15
Further Restrictions

„ We may impose further restrictions to the RNHD

„ For example, we may calibrate RNHD so that ATM

options are price at market levels

„ If there is value in the method, it should appear

without specific calibration on option prices

Page 16
Risk Neutralization

„ Solution: translate the historical density so that it is

centered on the forward

RNHD
3
Historical
Distribution
2

Log Variation

0
-40% -30% -20% -10% 0% 10% 20% 30% 40%

Forward
Page 17
RNHD

„ … And apply the variation density to the current index

level ...

4,0

3,5

3,0
Density on S is the
2,5
historical density of
Log(S/S0)
2,0

1,5

1,0

0,5
S0 (2440)
S
Index Level
0,0
1 500 1 700 1 900 2 100 2 300 2 500 2 700 2 900 3 100 3 300 3 500

-40% -30% -20% -10% 0% 10% 20% 30% 40%

S 
R = Log  
 S0  Page 18
RNHD

„ … to price any European contingent claim ...

4,0 2 000,0
RNHD 1 800,0
3,5
1 600,0
3,0
Call payoff 1 400,0
2,5
1 200,0

2,0 1 000,0

800,0
1,5
Call 600,0
1,0
Value 400,0
0,5 (x erT) 200,0
Index Level
0,0 0,0
1 500 1 700 1 900 2 100 2 300 2 500 2 700 2 900 3 100 3 300 3 500

Page 19
RNHD: Results

„ We compute the RNHD-implied skew in the last two

years and compare it to the market skew

30 September 2003
40%

35%

30%

25%

20%

€ Stoxx level
15%
1700 1900 2100 2300 2500 2700 2900 3100
RNHD Skew Implied skew Historical volatility

Page 20
 RNHD: Results

„ No improvement to the standard historical volatility

56% 8%

Moneyness = 100% 90% Moneyness Vol - ATM Vol

46% 6%
St Err = 7.0% Average = 4.1%

36% 4%

26% 2% Average = 0.7%

St Err = 9.2%

16% 0%

RNHD Historical
6% -2% RNHD Implied
Implied

-4% -4%
sept-01 mars-02 sept-02 mars-03 sept-03 sept-01 mars-02 sept-02 mars-03 sept-03

66% 56%
Moneyness = 80% Moneyness = 120%
56%
46%
St Err = 9.2%
46%
36%

36% St Err = 9.4%

26%
26%
St Err = 16% St Err = 8.4%
16%
16%
RNHD Historical
RNHD Historical 6%
6%
Implied Implied

-4% -4%
sept-01 mars-02 sept-02 mars-03 sept-03 sept-01 mars-02 sept-02 mars-03 sept-03

Page 21
Conclusion

„ Empirical application of RNHD is disappointing

Possible Reasons

„ The RNHD ignores Spot/Vol relationship. It provides

only sticky delta regimes

„ The RNHD ignores the daily nature of option hedging

„ The implementation is dependent on various

specifications (choice of kernel function and
bandwidth)

Page 22
Historical Skew

„ From the history, we compute the average hedged

P&L, assuming that the possible hedges are given by
B&S (σ is the only unknown)

Æ the option (strike K maturity T) is bought at an implied

volatility equal to σ, at date h

Æ it is subsequently hedged and valued using B&S delta

function at the σ level, every ∆t

„ The historical estimator will be such that the average

P&L is 0

„ By construction, the initial and final B&S valuation of

the option coincide with the actual cash-flows

Page 23
Historical Skew

„ P&L variation between t

and t+∆t
„ Hedge − δ (St , K ,σ ,T − t )∆St
1
θ (St , K ,σ ,T − t )∆t + δ (St , K ,σ ,T − t )∆St + γ (St , K ,σ ,T − t )∆St
2
„ Option
2
„ Financing δ (St , K ,σ ,T − t )St r∆t − C (St , K ,σ ,T − t )r∆t

„ Black & Scholes equation

1
θ + δSt r − Cr = − γSt2σ 2
2

1 
2  ∆St 
2

„ P&L variation γ (St , K ,σ ,T − t )St   − σ ∆t 
2

2  St  

„ accumulated P & L
T
∆t
−1
e r (T −i∆t )  ∆S 
2

P & L(σ , h) = ∑ γ (Sh + i∆t , K ,σ ,T − i∆t )Sh + i∆t  h + i∆t
2
 − σ ∆t 
2

i =0 2  Sh + i∆t  

Page 24
Historical Skew

„ For each period we compute the P&L of the hedged

position
2800
T
2600

2400

2200

2000

1800

The final return St+T / St do not matter. It’s the sequence of (∆St/St)2 that
drives the P&L.
Furthermore, the (∆St/St)2 will impact the P&L in proportion of the gamma of
the option of strike K at the point St

Page 25
Historical Skew

„ We may now calculate an historical average Profit and

Loss for the option over our data set, at the level σ

1 H
P & L(σ ) = ∑ P & L(σ , h )
H h =1

„ The fair implied volatility level for the option (strike K,

Maturity T) given the historical set is then given by:

σˆ K ,T solution of P & L(σ ) = 0

Page 26
Solving for the optimal σ

„ We introduce an “average gamma” function in the

Black & Scholes Greeks series:
T
1
γ (S, K ,σ ,T ) = ∫ e rT γ (S, K ,σ , t )dt
T 0

„ To get the equation defining σ (approximation):

1 H  ∆S  2 
0 = ∑ γ (Sh , K ,σ ,T )Sh 
2 h
 − σ ∆t 
2

h =1 2  Sh  

„ Equivalent expression:

γ (Sh , K ,σˆ K ,T ,T )Sh2

2
1 H
 ∆Sh 
σˆ K ,T =
∆t
∑ w 
h
 with w h = H
h =1  Sh 
∑ γ (S , K ,σˆ
i =1
i K ,T ,T )Si2

Page 27
Links with Black & Scholes

„ The historical skew is an extension of the historical

volatility with weights proportional to the gamma of
the option

„ In Black & Scholes: St and ∆St/St are independent

„ Thus we have:
2
1 2  ∆S 
P&L = γS   − σ ∆t
2

2  S 

„ Annulling the average P&L leads to the classic:

2
1 1  ∆Sh 
H
σ =
∆t
∑  
H h =1  Sh 

Page 28
Results

„ Application to the €-Stoxx 50 3-month options

As of 8 December 2003
35%

30%

25%

20%

15%

€ Stoxx level
10%
1800 2000 2200 2400 2600 2800 3000 3200 3400
Historical skew Implied skew Historical volatility

Page 29
 Results

„ The historical skew fits the implied better than

historical volatility

70% 8%
9m Historical skew
9m Historical Moneyness = 100%
Implied Skew
60% 9m Historical skew
6%
Implied StError = 5.3%
Average=4.1%
50%
4%

40%

2%
30%
Average=3.3%

0%
20%
août-01 févr-02 août-02 févr-03 août-03 févr-04
StError = 7%

10% -2%
août-01 févr-02 août-02 févr-03 août-03 févr-04

70% 70%

9m Historical Moneyness = 80% 9m Historical Moneyness = 120%

60% 9m Historical skew 60% 9m Historical skew
Implied Implied StError = 7.0%

50% 50%
StError = 5.1%

40% 40%
StError = 9.2%

30% 30%

20% 20%
StError = 9.3%

10% 10%
août-01 févr-02 août-02 févr-03 août-03 févr-04 août-01 févr-02 août-02 févr-03 août-03 févr-04

Page 30
Results

„ The historical Skew fits the €-Stoxx 50 option implied

much better than the RNHD

Quadratic distance to the implied volatility

18%

16% RNHD
Historical Vol
14%
Historical Skew
12%

10%

8%

6%

4%

2%

0%
80% 90% 100% 110% 120%

Page 31
Application to other derivatives

„ This approach may be extended to various sorts of

equity derivatives

„ We consider the family of “Black & Sholes” replicating

strategies (Delta & Gamma functions of the spot)

„ It is then possible to determine the volatility parameter

such that the average hedged P&L is 0

„ We thus have a direct historical pricing of the

derivative

Page 32
Conclusion

„ The historical skew introduced is a natural extension

of the historical volatility that takes strike and maturity
into account

„ It allows to construct a “fair skew” based solely on

historical returns, without any pricing model

„ It relies on the selection of the B&S hedge that is

neutral on historical average

„ It improves the accuracy of the historical estimator,

even for ATM options

„ It outperforms the RNHD estimator

Page 33
Disclaimer

In accordance with US regulations, the analyst(s) named in this report certifies that 1) all of the views expressed herein accurately reflect the personal views of
the analyst(s) with regard to any and all of the content, securities and issuers and 2) no part of the compensation of the analyst(s) was, is, or will be, directly or
indirectly, related to the specific recommendation or views expressed by the analyst(s) in this report.
It is a BNP Paribas policy that analysts are prohibited from trading in any of the stocks in the sector that they cover.
THIS PUBLICATION IS NOT INTENDED FOR PRIVATE CUSTOMERS AS DEFINED IN THE FSA RULES AND SHOULD NOT BE PASSED ON TO ANY
SUCH PERSON.
The material in this report was produced by a BNP Paribas Group Company. It will have been approved for publication and distribution in the United Kingdom by
BNP Paribas London Branch, a branch of BNP Paribas SA whose head office is in Paris, France. BNP Paribas London Branch is regulated by the Financial
Services Authority ("FSA") for the conduct of its designated investment business in the United Kingdom and is a member of the London Stock Exchange. BNP
Paribas Securities Corp. in the United States accepts responsibility for the contents of this report in circumstances where the report has been distributed by BNP
Paribas Securities Corp. direct to U.S. recipients.
This report is being distributed to Japanese based firms by BNP Paribas Securities (Japan) Limited, Tokyo Branch, or by a subsidiary or affiliate of BNP Paribas
not registered as a securities firm in Japan, to certain financial institutions permitted by regulation. BNP Paribas Securities (Japan) Limited, Tokyo Branch, a
subsidiary of BNP Paribas, is a securities firm registered according to the Securities & Exchange Law of Japan and a member of the Japan Securities Dealers
Association. BNP Paribas Securities (Japan) Limited, Tokyo Branch accepts responsibility for the content of a report prepared by another non-Japan affiliate
only when distributed to Japanese based firms by BNP Paribas Securities (Japan) Limited, Tokyo Branch.
The information and opinions contained in this report have been obtained from public sources believed to be reliable, but no representation or warranty, express
or implied, is made that such information is accurate or complete and it should not be relied upon as such. Information and opinions contained in the report are
published for the assistance of recipients, but are not to be relied upon as authoritative or taken in substitution for the exercise of judgement by any recipient, and
are subject to change without notice. This report is not, and should not be construed as, an offer document or an offer or solicitation to buy or sell any
investments. Any reference to past performance should not be taken as an indication of future performance. No BNP Paribas Group Company accepts any
liability whatsoever for any direct or consequential loss arising from any use of material contained in this report. This report is confidential and is submitted to
selected recipients only. It may not be reproduced (in whole or in part) to any other person. A BNP Paribas Group Company and/or persons connected with it
may effect or have effected a transaction for their own account in the investments referred to in the material contained in this report or any related investment
before the material is published to any BNP Paribas Group Company's customers. On the date of this report a BNP Paribas Group Company, persons connected
with it and their respective directors and/or representatives and/or employees may have a long or short position in any of the investments mentioned in this report
and may purchase and/or sell the investments at any time in the open market or otherwise, in each case either as principal or as agent. Additionally, a BNP
Paribas Group Company within the previous twelve months may have acted as an investment banker or may have provided significant advice or investment
services to the companies or in relation to the investment(s) mentioned in this report.
This report is prepared for professional investors and is not intended for Private Customers in the United Kingdom as defined in FSA rules and should not be
passed on to any such persons. For the purpose of distribution in the United States this report is only intended for persons which can be defined as 'Major
Institutional Investors' under U.S. regulations. Any U.S. person receiving this report and wishing to effect a transaction in any security discussed herein, must do
so through a U.S. registered broker dealer. BNP Paribas Securities Corp. is a U.S. registered broker dealer.
By accepting this document you agree to be bound by the foregoing limitations.

© BNP Paribas (2003). All rights reserved.

Page 34