Equity Derivatives & Quantitative Research

19th February 2004

Investing in Implied Volatility

Andrew Harmstone Introduction

+44 207 103-4275
aharmsto@lehman.com Volatility tends to be negatively correlated with equity returns – especially at the
index level. Moreover volatility appears to display mean reversion and other
forecastable properties.1 In view of this, investors might wish to invest in
volatility to hedge their equity investments or to take directional positions.

The classic investment in volatility has been a delta hedged straddle or more
recently a volatility/variance swap. Although useful, the payoff of both of these
instruments has the disadvantage of being tied to the difference between realized
volatility and implied volatility – not the level of implied volatility itself. More
recently volatility indices have become popular. These have been successful in
identifying trends in volatility, but it has not been easy to design products that
represent direct investments in the indices.

This report begins with a short review of the popular approaches to estimating
implied volatility. It reviews the theory behind recent implied volatility indices,
then it shows why investing directly in these indices has proven a challenge.
Finally, it ends with possible investible products.

Popular Approaches Estimating Implied Volatility

Figure 1: Implied Volatility Report on European Indices for 3rd February 2004
Change in 3- Fixed Strike: Change in
Latest 3-Month Implied
Index 1 Day Return Month ATM 3-Month Implied
Price Volatility
Implied Volatility Volatility

FTSE 4390.6 0.2% 14.5% -0.28↓ -0.16↓

ESTOXX 2841.26 -0.4% 18.7% 0.32↑ 0.08↑
DAX 4057.51 -0.3% 21.5% 0.03↑ 0.02↑
CAC 3638.21 -0.7% 18.4% 0.32↑ 0.00↔
MIB30 27615 -0.6% 15.1% 0.63↑ 0.42↑
SMI 5735 -0.9% 15.6% 0.01↑ 0.00↔
AEX 351.74 -1.0% 19.9% 0.89↑ 0.34↑
S&P 500 1136.03 0.1% 14.6% -0.03↓ 0.00↔
Nikkei 10641.92 -1.3% 22.4% -0.18↓ -0.33↓
Source: Lehman Brothers, Bloomberg

1
Volatility as an Asset Class, Lehman Brothers 2002

PLEASE SEE ANALYST CERTIFICATION AND IMPORTANT DISCLOSURES BEGINNING ON PAGE 9.

INITIAL VERSION OF THIS REPORT HAS BEEN CIRCULATED PRIOR TO 19th FEBRUARY 2004
Equity Derivatives & Quantitative Research

Ever since the Black Scholes options pricing model was developed it has been
possible to estimate the level of volatility a stock must have in order to justify its
current price. This estimated volatility is generally called implied volatility. This
is still the most widely used way of measuring volatility. Figure 1 shows a
volatility report on the European indices.

Figure 1 shows what is called the At-The-Money (ATM) implied volatility for
each index for a term of three months ahead. This is calculated by choosing the
options for each index which have strike prices close to the spot price of the
option’s underlying index and then interpolating. Unfortunately, this estimate is
not stable in the sense that the implied volatility calculated from an option with a
different strike price could be meaningfully different from that calculated from
the ATM option.

Figure 1 shows that the estimated change in volatility can be significantly

influenced by which strike price is chosen. For example, the ATM implied
volatility for the CAC 40 index over a three-month horizon is 18.2%. It rose by
0.32% from 2nd February 2004 to the 3rd February 2004. But this result holds only
if the strike is used for both days is ATM.2 Alternatively, it is equally reasonable
to use the same strike option on both days. If this is done then Figure 1 shows that
the implied volatility for the CAC 40 over a three month horizon was unchanged
from 2nd February 2004 to the 3rd February 2004. This is shown in the column
labelled “Fixed Strike: Change in 3-Month Implied Volatility”. As it turns out the
CAC 40 fell by 0.7% on the 2nd February 2004 and because of this the ATM
strike on the 3rd February was below that on the 2nd. Arguably, the rise in the
ATM volatility between the 2nd and 3rd of February was an illusion caused by the
market decline making the ATM strike fall. This is because lower strike options
generally have higher implied volatility.

2
The volatilities in Figure 1 are determined from an estimated implied volatility surface that is
determined from actual option prices. This makes it possible to estimate the ATM volatility even
when the index level is not at a particular listed option’s strike price.

2
Equity Derivatives & Quantitative Research

Figure 2: 3M Skew for the EuroSTOXX 50. 100% strike is ATM

35%

30%

25%

Implieds 20%

15%

10%
80% 90% 100% 105% 110% 120%

Strike

Source: Lehman Brothers, Bloomberg

Figure 2, using the EuroSTOXX 50 index as an example, shows the difference in

volatility for different strike prices can be substantial – this is the “skew” in the
implied volatility surface. The skew makes the implied volatility calculation from
an option pricing model like the Black Scholes model unsatisfactory. First of all,
there is the obvious contradiction of estimating volatility that changes over
different strike prices with a model that explicitly assumes that volatility remains
constant. Second, direct investment in implied volatility calculated by the model
is difficult. This is because the implied volatility is based on the ATM volatility,
but the ATM option changes every day as markets move. This means that an
investment product tied to the implied volatility defined in this way must
continuously rebalance to the current ATM option.

Volatility Indices

Pioneering volatility indices were created by the CBOE and by the Deutsche
Börse, among others. These include the “old” VIX® and the VDAX®. Both of
these indices are based on average implied volatility levels of options that are
close to being at-the-money. These indices were successful in capturing, at least
in a broad sense, the trend in implied volatility. Direct investment in these indices
was hampered, however, by the fact that no one option or portfolio of options
would track the changes in implied volatility that occurred as the level of the
market changed leading to changes in the ATM strike. Thus even though futures
existed on both indices, market makers found it hard to hedge their positions in
the futures and liquidity stayed low.

Simultaneous with the creation of volatility indices, there developed an Over-the-

Counter (OTC) market in volatility and variance swaps. These were

3
Equity Derivatives & Quantitative Research

generalizations of delta hedged straddles. A variance swap, for example, can be

constructed from a portfolio of options that behaves like a “generalized” straddle.
Like straddles these options portfolios have “positive” gamma so that delta
hedging is generally profitable. That is, if markets rise, the delta of the options
tends to rise, leading to sales of futures to reduce the delta down to zero. Similarly
if markets fall, then a positive gamma causes the options delta to drop, leading to
purchases of futures to bring the delta back to its target of zero exposure. Note
this strategy is inherently profitable because it means the delta hedger is “selling
high” and “buying low”. Moreover, the more volatile the markets are, the more
delta hedging occurs, and the more profit there is from this activity.

The implication is that the higher realized volatility is, the more profitable the
variance swap will be. But, of course, there is no free lunch because any option
position with a positive gamma will also have a negative “theta”. That is, it will
decline in value over time. It turns out that the rate of decline over time is a
function of the option portfolio’s “implied volatility”. Therefore, the net profit
from a variance swap depends on two offsetting factors. The first is the level of
realized volatility. The second factor is the option portfolio’s implied volatility
was when it was initiated. Clearly the variance swap payoff is tied to the
difference between implied volatility and realized volatility. Conceptually, it
“converts” the implied volatility embedded in the options portfolio into realized
volatility generated from delta hedging as the swap goes from inception to
expiration.

Finally, the main advantage of using the portfolio of options in a “generalized

straddle” over using a traditional straddle consisting of a long call and a long put
is that the gamma of the generalized straddle is relatively stable when market
prices change. This is clearly not the case with a simple straddle because, if, for
example, prices rise then the call side of the straddle becomes in the money
rapidly and its delta rises up to a maximum of 1 and its gamma falls. But with a
lower gamma there is less profitable delta hedging. So a straddle rapidly loses its
payoff from high realized volatility in either an upward trending or a downward
trending market. The variance swap options portfolio is substantially insulated
from this effect because it is explicitly constructed to keep the gamma stable for a
wide range of price levels. So even in up or down markets it retains the ability to
convert realized volatility into profit. Pari passu keeping gamma constant means
its rate of decline over time, or negative theta, stays stable as well. This means
that, unlike a straddle, a variance swap payoff, tied to the delta-hedged
generalized straddle, maintains its desired trade-off between realized volatility
and implied volatility even if market prices move substantially.

Clearly the value of the options portfolio used in a variance swap is a function of
implied volatility (technically implied variance). Therefore, it seems natural to use
its value as a proxy for the level of implied variance. One of the most remarkable
features of the variance swap option portfolio is that its value at inception (with a
minor adjustment) is in fact exactly equal to a very reasonable estimate of implied
volatility. This mathematical fact forms the basis of the most recent generation of
implied volatility indices.

4
Equity Derivatives & Quantitative Research

Latest Generation of Volatility Indices

The new VIX® announced by the CBOE September 2003 is an example of the
latest generation of implied volatility indices. It is an important advance over the
prior generation because it does not use any option pricing model to determine the
level of implied volatility.3 It provides a single summary estimate of volatility
across the whole volatility surface in a natural way. Finally, it summarizes the
characteristics of the whole volatility surface instead of being tied to ATM or near
ATM strike options. It tends to be more correlated with the performance of
volatility sensitive investment strategies. It accomplishes this because it is based,
in essence, on the value of the generalized straddle used typically in constructing
a variance swap, as discussed above. Note, however, that by nature this index and
any other index based on the same methodology is fundamentally tied to implied
variance, not implied volatility. The volatility can always be calculated from the
variance, but the movement in volatility will never be linear with respect to the
movement of the underlying portfolio of options that drive the index value.

The underlying principle of the new indices is discussed in the next section; a
more detailed derivation is available on request.

The Key is Reversing the Order of Summation (Integration)!

The “generalized straddle” is essentially a portfolio of out-of-the money call

options and out-of the-money put options. The call component, with strikes, Kj,
CallPortfolio, is

CallPortfolio = ΣFor all Kj>= Km (∆K / Kj2) OTMCj.

Km is the lowest strike price that is still higher than the forward price F. ∆K is the
difference between each strike price and for simplicity we assume that it is
constant.

The current value of a call can be calculated as the probability weighted average
of the call’s end of period value. Figure 3 illustrates the probability distribution
and the corresponding end of period values.

Suppose that out of a discrete set of end of period stock prices from 0 to some
large number, Sn, the highest discrete stock price that is less than F is Sg. Next,
suppose that the probability that the end of period stock price is Sg is pg. Then the
value of the j’th OTM call with strike price Kj, OTMCj, can be written as

(1) OTMCj = Σi=g+1 to n ρpi * Max(0, Si – Kj) where Kj > F, the futures price,

3
A few assumptions about the stock return distribution cannot, however, be avoided. The most
important may be the assumption that the distribution has a finite variance.

5
Equity Derivatives & Quantitative Research

where ρ is the present value discount factor. This means that the call side of the
generalized straddle is

Figure 3: Estimating the Probability Weighted Value of the OTM 110 Call

Cross-Section: Holding Strike Constant

Prob Wgted Ending Option Value is Current Option Price!

50.0 2.5%
OTM Call
40.0 2.0%

Call Value at
Stock Prob

Probability
Expiration
110
30.0 1.5%
20.0 1.0%
10.0 0.5%
0.0 0.0%

0
0

0
50

70

90
60

80

10

11

12

13
14

15
Ending Stock Price

Source: Lehman Brothers

(2A) CallPortfolio = ΣFor all Kj>= Km (∆K / Kj2) {Σi=g+1 to n ρpi *Max(0, Si – Kj)}.

Similarly, the put component of the generalized straddle is,

(2B) PutPortfolio = ΣFor all Kj < Km (∆K / Kj2) {Σi=1 to g ρpi *Max(0, Kj – Si)}.

The key step is reversing the order of summation in equations (2A) and (2B). This
changes the right hand side of each equation from a portfolio of options into an
expression that includes the expected log of the normalized end of period stock
price. The reason this is critical is that this expected log is linearly related to the
variance of the stock price.

To see that this is the case, first examine the equations after the order of
summation has been reversed:

(3A) CallPortfolio = Σi=g+1 to n ρpi *{ ΣFor all Kj>= Km (∆K / Kj2) *Max(0, Si – Kj)}.

(3B) PutPortfolio = Σi=1 to g ρpi * {ΣFor all Kj < Km (∆K / Kj2) *Max(0, Kj – Si)}.

The generalized straddle, GenSD, then is

GenSD = Σi=1 to n ρpi * {ΣFor all Kj>= Km (∆K / Kj2) *Max(0, Si – Kj)
+ ΣFor all Kj < Km (∆K / Kj2) *Max(0, Kj – Si) }.

But it can be shown that the term in brackets approximates

6
Equity Derivatives & Quantitative Research

S i /F – 1 - Log (S i /F).4

This means that the generalized straddle approximates,

5
Σi=1 to n ρpi * { S i /F – 1 – Log (S i /F)}, which simplifies to ,

Σi=1 to n ρpi * { – Log (S i /F)} or, approximately, (-ρ) * Expected Log(S i /F).

That is the generalized straddle equals the negative of the discounted expected log
of the ending stock price normalized by the current futures price.

It is well-known that the expected log itself equals,

Expected Log (S i /F) = Tr – Log(F/ S0) - T/2 Variance.

Here, Variance is the annualized variance of the stock price, T is the time to
expiration of the options (as a % of the year) and r is the annualized risk free rate.

Combining these results shows that 6

GenSD = CallPortfolio + PutPortfolio

~ ρ T/2 Variance.

In the case where the strike price, Km, does not equal the current futures price an
additional term is required and the relationship becomes,

GenSD = CallPortfolio + PutPortfolio ~ ρ T/2 Variance + ρ ½ (F / Km- 1)2,

where ~ means “approximates”. Solving for the Variance,

Variance ~ (1/ρ) (2/T) (CallPortfolio + PutPortfolio) - 1/T (F / Km- 1)2.

To get a constant maturity variance it is possible to interpolate the value of the

GenSD for two options series chosen such that the weighted average time to
expiration matches the desired constant maturity. Next, a current generation
volatility index annualizes the variance and takes its square root to calculate the
desired annualized volatility.

4
There is another term that reflects the fact that Km does not exactly equal the futures price, F. The
value of this term can be approximated by ½ ρ (F / Km- 1)2
5
The expected value of the stock price, (approximately equal to Σi=1 to n pi * { S i }), equals the
futures price because by construction the probability distribution is “risk-neutral”. This causes the
first two terms in the summation to cancel out.
6
Tr - Log(F/S0) is close to zero. For example, if there are no dividends and there is a flat risk free
rate then, F = (1+r)T S0, making Log(F/S0) = Tlog(1+r). Since Tr is the first term of a Taylors
expansion for Log(1+r), Tlog(1+r) is close to Tr. The difference is of the order of the square of the
risk free rate.

7
Equity Derivatives & Quantitative Research

Conclusion

The latest generation of volatility indices is based on the value of an options

portfolio that behaves like a generalized straddle. The significant innovation is
that the generalized straddle is constructed so that for a wide range of prices for
the underlying, the “gamma” of the options portfolio is stable. The remarkable
mathematical result is that the value of this generalized straddle, which can be
calculated solely from the market prices of the options in the portfolio,
approximates the implied volatility of the underlying.

Nonetheless, the implied variance estimated by the volatility index is not directly
investable. For example, it may seem natural simply to buy the options portfolio
that underlies the index. But one day after this portfolio is purchased, the options
experience time decay – even if the implied volatility does not drop. Moreover,
the options also have a delta and so the value of the options portfolio can rise and
fall with the market even if volatility levels do not change. If delta hedging is
introduced then realized volatility impacts the payoff. So the payoff to a product
that attempts to tie itself directly to an implied volatility index has a payoff that is
a hybrid of the changes in implied volatility and other factors such as market
movements or realized volatility combined with decay due to the passage of time.

The reason for this apparent paradox is that implied volatility is strictly a forward
looking concept. One day after it is estimated, implied volatility is “corrected” by
the actual volatility that occurs in the market place. Implied volatility can be
compared to the energy stored in a boulder at the top of a hill. Initially, all the
energy is potential (corresponding to implied volatility). Once the boulder starts
down the hill its energy changes from potential energy to kinetic energy (like
realized volatility). If there is no friction and no force is applied to the boulder
then the total energy level remains constant so that at the bottom of the hill the
total kinetic energy will be exactly equal to the initial potential energy. This
would correspond to the case where realized volatility was exactly what was
predicted by the implied volatility. But, in reality, a rolling boulder experiences
friction and other forces that affect its energy level so the translation from
potential energy to kinetic is not fully predictable and the total energy in the end
may differ substantially from the potential energy at the top of the hill. Similarly,
the realized volatility will differ from the implied.

Continuing with the analogy, it is not appropriate to estimate the total energy level
of a moving boulder by just calculating its potential energy. Similarly, it is not
appropriate to estimate the volatility of a stock index by looking only at implied
volatility once the index starts moving over time. The only time implied volatility
truly corresponds to the total volatility of the stock or stock index is before it
“starts moving”. That is to say, a product on implied volatility should be to be
forward looking, delivering the implied volatility as of some future date.

8
CURRENT RESEARCH DISCLOSURES, DISTRIBUTION OF OUR RATINGS, AND PRICE CHARTS REGARDING COMPANIES
MENTIONED IN THIS DOCUMENT MAY BE OBTAINED BY GOING TO : THE LEHMAN BROTHERS WEBSITE:
HTTP://WWW.LEHMAN.COM/DISCLOSURES
OR BY SENDING A WRITTEN REQUEST REFERENCING THE TITLE AND DATE OF THIS DOCUMENT TO : LEHMAN
BROTHERS INC. RESEARCH DISCLOSURES, 745 7TH AVENUE, 8TH FLOOR, ATTENTION: CONTROL ROOM, NEW YORK,
NY 10019

Analyst Certification:
I, Andrew Harmstone, hereby certify (1) that the views expressed in this research report accurately reflect my personal views about any or
all of the subject securities or issuers referred to in this report and (2) no part of my compensation was, is or will be directly or indirectly
related to the specific recommendations or views contained in this report.

To the extent that any of the views expressed in this research report are based on the firm's quantitative research model, Lehman Brothers
hereby certify (1) that the views expressed in this research report accurately reflect the firm's quantitative research model and (2) that no
part of the firm's compensation was, is or will be directly or indirectly related to the specific recommendations or views expressed in this
report.

Important Disclosures:
Disclosures are made as of the last trading day of the previous month.
The analysts responsible for preparing this report have received compensation based upon various factors including the Firm's total
revenues, a portion of which is generated by investment banking activities.
Please note that the trade ideas within this report in no way relate to the fundamental ratings applied to European stocks by Lehman
Brothers' Equity Research.

Key to Investment Opinions:

Stock Rating
1 - Overweight - The stock is expected to outperform the unweighted expected total return of the industry sector over a 12-month
investment horizon.
2 - Equal weight - The stock is expected to perform in line with the unweighted expected total return of the industry sector over a 12-month
investment horizon.
3 - Underweight - The stock is expected to underperform the unweighted expected total return of the industry sector over a 12-month
investment horizon.
RS - Rating Suspended - The rating and target price have been suspended temporarily to comply with applicable regulations and/or firm
policies in certain circumstances including when Lehman Brothers is acting in an advisory capacity in a merger or strategic transaction
involving the company.

Sector View
1 - Positive - sector fundamentals/valuations are improving
2 - Neutral - sector fundamentals/valuations are steady, neither improving nor deteriorating
3 - Negative - sector fundamentals/valuations are deteriorating
Stock Ratings From February 2001 to August 5, 2002 (sector view did not exist):
This is a guide to expected total return (price performance plus dividend) relative to the total return of the stock’s local market over the next
12 months.
1=Strong Buy - expected to outperform the market by 15 or more percentage point.
2=Buy - expected to outperform the market by 5-15 percentage points
3=Market Perform - expected to perform in line with the market, plus or minus 5 percentage points
4=Market Underperform - expected to underperform the market by 5-15 percentage points
5=Sell - expected to underperform the market by 15 or more percentage points.

Distribution of Ratings:

Lehman Brothers Equity Research has 1612 companies under coverage.

40% have been assigned a 1-Overweight rating which, for purposes of mandatory regulatory disclosures, is classified as a Buy rating. 39%
of companies with this rating are investment banking clients of the Firm.
40% have been assigned a 2-Equal weight rating which, for purposes of mandatory regulatory disclosures, is classified as a Hold rating,
10% of companies with this rating are investment banking clients of the Firm.
20% have been assigned a 3-Underweight rating which, for purposes of mandatory regulatory disclosures, is classified as a Sell rating, 66%
of companies with this rating are investment banking clients of the Firm.

Lehman Brothers produces a number of different types of research product including, amongst others, fundamental analysis, quantitative analysis and
short term trading ideas. Recommendations contained in one type of research product may differ from recommendations contained in other types of
research product, whether as a result of differing time horizons, methodologies, or otherwise. Should you wish to receive any research product of a
type that you do not presently receive, please contact your Lehman Brothers sales representative who will be pleased to assist you.
This material has been prepared and/or issued by Lehman Brothers Inc., member SIPC, and/or one of its affiliates ("Lehman Brothers") and has been
approved by Lehman Brothers International (Europe),authorised and regulated by the Financial Services Authority, in connection with its distribution in
the European Economic Area. This material is distributed in Japan by Lehman Brothers Japan Inc., and in Hong Kong by Lehman Brothers Asia
Limited. This material is distributed in Australia by Lehman Brothers Australia Pty Limited, and in Singapore by Lehman Brothers Inc., Singapore
Branch. This material is distributed in Korea by Lehman Brothers International (Europe) Seoul Branch. This document is for information purposes only
and it should not be regarded as an offer to sell or as a solicitation of an offer to buy the securities or other instruments mentioned in it. No part of this
document may be reproduced in any manner without the written permission of Lehman Brothers. We do not represent that this information, including
any third party information, is accurate or complete and it should not be relied upon as such. It is provided with the understanding that Lehman
Brothers is not acting in a fiduciary capacity. Opinions expressed herein reflect the opinion of Lehman Brothers and are subject to change without
notice. The products mentioned in this document may not be eligible for sale in some states or countries, and they may not be suitable for all types of
investors. If an investor has any doubts about product suitability, he should consult his Lehman Brothers representative. The value of and the income
produced by products may fluctuate, so that an investor may get back less than he invested. Value and income may be adversely affected by
exchange rates, interest rates, or other factors. Past performance is not necessarily indicative of future results. If a product is income producing, part of
the capital invested may be used to pay that income. Lehman Brothers may, from time to time, perform investment banking or other services for, or
solicit investment banking or other business from any company mentioned in this document. © 2004 Lehman Brothers. All rights reserved. Additional
information is available on request. Please contact a Lehman Brothers’ entity in your home jurisdiction.
Complete disclosure information on companies covered by Lehman Brothers Analysts is available at www.lehman.com/disclosures