Equity Research

28 April 2014

Special Report
DERIVATIVES
Market Neutral Variance Swap & VIX
U.S. Equity Derivatives Strategy
Futures Strategies
U.S. Equity Derivatives Strategy
It is perhaps not well appreciated that short variance swap (SVS) strategies which Maneesh S. Deshpande
efficiently capture the volatility risk premium (VRP) have a significant market 1.212.526.2953
maneesh.deshpande@barclays.com
exposure. Removing this market exposure requires an additional position in the
BCI, New York
underlying index. In this note, we present several methods to calculate this
variance swap delta and thus construct market-neutral SVS strategies. In general, Ashish Goyal
we find that the Sharpe ratios of these strategies do not materially change, attesting 1.212.526.2771
to the fact that the VRP cannot be fully attributed to the Equity Risk Premium (ERP). ashish.x.goyal@barclays.com
BCI, New York
We first use the well-known replication strategy for variance swaps to derive an
Arnab Sen
explicit formula for the variance swap delta. We show that a non-zero value requires
1.212.526.5429
the presence of a volatility smile. Our first method to calculate the variance swap arnab.sen@barclays.com
delta is to simply use the Black-Scholes formula for the delta of the replicating BCI, New York
option strip. This is equivalent to the sticky strike model, where strike volatilities
are assumed to not change. This approach does not capture the fact that strike
volatilities themselves react and hence does not fully remove the market exposure.

We next show that the variance swap delta can also be calculated by simply
combining the sticky strike assumption with the approximation formula for the
variance swap strike as a function of smile curve parameters. The advantage of this
approach is that we can now incorporate the empirical dynamics of the strike
volatilities by using the skew stickiness ratio (SSR), which can be calculated by
regressing changes in at-the-money (ATM) volatility versus the skew-adjusted
index returns. Historically, the SSR is ~1.5 as opposed to the value of 1 predicted by
the sticky strike model.

A simple way to incorporate this empirical fact is to simply scale up our calculated
sticky-strike variance swap delta by the empirical SSR. We show that the empirical
market beta of the resulting hedged SVS strategy is now almost zero.

In the final section, we use the above methodology to calculate the delta of VIX
futures and construct market neutral-short VIX futures strategies. The results are
similar in that the Sharpe ratios are not materially affected by removing the market
beta.

Barclays Capital Inc. and/or one of its affiliates does and seeks to do business with companies
covered in its research reports. As a result, investors should be aware that the firm may have a
conflict of interest that could affect the objectivity of this report.
Investors should consider this report as only a single factor in making their investment decision.
PLEASE SEE ANALYST CERTIFICATION(S) AND IMPORTANT DISCLOSURES BEGINNING ON PAGE 19.
Barclays | Special Report

Introduction
It is perhaps not well appreciated that short variance swap (SVS) strategies that efficiently
capture the volatility risk premium have a significant market exposure. This beta stems from
the fact that realized volatility itself is negatively correlated with market returns. We
estimate that nearly 25% of the return of a SVS strategy can be attributed to its beta
exposure. Removing this market exposure requires an additional position in the underlying
index, which we define to be the delta of the variance swap. In this note, we present
several methods to calculate this delta and thus construct market neutral SVS strategies.

Note that this variance swap delta is quite distinct from the exposure required for replicating
the variance swap. The well known replication strategy for variance swaps involves a static
(i.e. non delta-hedged) position in a strip of options and a dynamic exposure to the
underlying (the replication delta), which is completely model independent. In a pure
Black-Scholes world, the delta of the option strip exactly cancels the replication delta and
thus the variance swap delta is always zero.

In the presence of a volatility smile, the variance swap delta is not zero. Our first method to
calculate the variance swap delta is to simply use Black-Scholes formula for the option strip
delta. This approach implicitly assumes a sticky strike model for the implied volatility
surface where the strike volatilities are assumed to stay constant when the underlying
moves, which of course is not true empirically. Indeed, we show that with this approach,
although the market exposure of the SVS strategy decreases, it does not drop completely to
zero.

We next show that the option strip delta can also be calculated by simply combining the
sticky strike assumption with the approximation formula for the variance swap strike as a
function of smile curve parameters. The advantage of this approach is that we can now
incorporate the empirical dynamics of the strike volatilities by using the skew stickiness
ratio (SSR), which can be calculated by regressing changes in ATM volatility versus the
skew adjusted index returns. Empirically, the SSR is ~1.5 as opposed to the value of 1
predicted by the sticky strike model. A simple way to incorporate this empirical fact is to
simply scale up our calculated variance swap delta by the empirical SSR. We show that the
empirical market beta of the resulting SVS strategy is now almost zero. In general, we find
that the Sharpe ratio of this strategy does not materially change, attesting to the fact the
VRP cannot be fully attributed to the Equity Risk Premium (ERP).

The results of this note also help in understanding the difference in performance of delta
hedged straddles with that of a pure variance swap, especially during periods of strong
market rallies. While the DH straddles suffer from more path dependency, they also have
less positive market exposure, because of which they are likely to underperform un-hedged
variance swaps during strong market rallies. In contrast, delta-hedged straddles perform
much better during market downturns.

In the final section, we use the above methodology to calculate the delta of VIX futures and
construct market neutral short VIX futures strategies. The results are similar in that the
Sharpe ratios are not materially affected by removing the market beta.

28 April 2014 2
Barclays | Special Report

Variance Swap Strategies: Pure but not market neutral

Variance swaps are a pure means to trade volatility since their payoff is simply the
difference in the squares of implied and realized volatility. Thus the payoff for a variance
swap with variance notional of $1 can be written as:

Where is the variance swap strike and is the realized volatility during the tenor of the
variance swap. In contrast, a delta-hedged option strategy suffers from significant path
dependence in the sense that paths with the same realized volatility can have significantly
different payoffs. This feature has made short variance swap (SVS) strategies an attractive
vehicle for investors to harvest the premium of option implied volatility versus subsequent
realized volatility (the so called Volatility Risk Premium, or VRP). As shown in Figure 1, long
term performance of SVS strategies is quite attractive.

FIGURE 1 FIGURE 2
Short variance swap strategies have outperformed With attractive risk-adjusted returns
equities

Cum.ExcessReturns
Sharpe Draw-
350 Strategy Avg. Ret. Std. Dev.
Ratio down

300

250 SPX ER 7.2% 20.1% 0.36 -61%

200

150
SVS 6.5% 5.2% 1.25 -24%
100

50
Jan-96 Jan-00 Jan-04 Jan-08 Jan-12
SPX SVS
Source: Barclays Research, Bloomberg, OptionMetrics Source: Barclays Research, OptionMetrics, Bloomberg
Note: Excess return for SPX calculated by subtracting Fed-fund rates from the Note: Excess return for SPX calculated by subtracting Fed-fund rates from the
daily returns of SPX total returns. The SVS strategy sells 0.2 vega of 1M variance daily returns of SPX total returns. The SVS strategy sells 0.2 vega of 1M variance
swap for $100 of capital. Transaction costs are ignored. swap for $100 of capital. Transaction costs are ignored. Data from Jan 1996 to
April 2014.

However, as can been seen visually from Figure 1, the SVS strategy appears to be correlated
with the underlying index. Quantitatively, the daily and monthly correlations between the
two indices are 58% and 55%, respectively. Figure 3 shows that this market dependence is
quite non-linear. As expected, large negative index returns will naturally result in large
draw-downs for the SVS strategy. However, it is noteworthy that the slope around the origin
or the beta for small market moves is still non-zero.

At first glance, this market dependence might appear to be at odds with the fact that the
P&L of a variance swap is supposed to only depend on realized volatility and not the
underlying index return. After all, isnt the whole point of using variance swaps to mitigate
the path dependence of the payoff?

28 April 2014 3
Barclays | Special Report

FIGURE 3 FIGURE 4
Short variance strategy has a non-linear market exposure Driven by the strong asymmetric dependence of realized
volatility on index returns

SVS 1M Ret SPX 1M RV

5% 100
90
0%
80
-5% 70
60
-10% 50
40
-15%
30
-20% 20
10
-25% 0
-25% -20% -15% -10% -5% 0% 5% 10% 15% 20% -25% -20% -15% -10% -5% 0% 5% 10% 15% 20%
SPX 1M Ret SPX 1M Ret
Source: Barclays Research, OptionMetrics, Bloomberg Source: Barclays Research, OptionMetrics, Bloomberg
Note: Dashed lines denote the 1 std deviation band around the mean. The SVS Note: Dashed lines denote the 1 std deviation band around the mean
strategy sells 0.2 vega of 1M variance swap for $100 of capital. Transaction costs
are ignored. Data from Jan 1996 to April 2014.

The resolution of this apparent contradiction is a bit subtle. The P&L of a variance swap is
indeed only dependent on the realized volatility of the path and not its other details
(including the total return). However, it is an empirical fact that realized volatility is higher
when the market sells off (as shown in Figure 4). Although the implied volatility (or the
variance swap strike) does try its best to capture the distribution of realized volatility, it is
after all a single number, and as a result the dependence of realized volatility on market
returns means that some market exposure is unavoidable for a SVS strategy.

Note that for an opportunistic investor who trades variance swaps to express a view on
subsequent realized volatility, there is no contradiction. As long as their view of realized
volatility is correct, the P&L of the variance swap will be as expected. However, and this is
the crucial point, any forecast of realized volatility also has an imbedded implicit market
return forecast.

The above discussion implies that at least part of the VRP harvested using SVS strategies is
simply a result of the fact that equities have rallied, or that an Equity Risk Premium (ERP)
exists. To quantitatively assess this, we regress the SVS strategy versus SPX returns and
look at the intercept (alpha). Figure 5 shows the results for a SVS strategy where we sell 0.2
vega for $100 capital. Since the average return of the SVS over this time period was 0.51%
per month, we see that 0.12% can be explained by market beta. Said differently, on average
2.5 vol points were made over this time period using variance swaps, but 0.6 volatility point
can be attributed to the positive market return over this period.

28 April 2014 4
Barclays | Special Report

FIGURE 5
About 25% of the returns of SVS can be attributed to market exposure

Avg. SVS Monthly

Alpha Beta to SPX Alpha tStat Beta tStat R^2
Ret.

0.51% 0.39% 0.18 2.05 2.16 30.2%

Source: Barclays Research

Note: Monthly SVS returns are regressed against monthly SPX returns. T-stats adjusted for overlapping returns. The
SVS strategy sells 0.2 vega of 1M variance swap for $100 of capital. No transaction costs assumed. Data from Jan 1996
to April 2014.

Thus an investor who would like a market neutral SVS strategy would need to hold an
additional short position in the underlying index, which we denote as the variance swap
delta. The main goal of the rest of this note is to develop a methodology to calculate this
variance swap delta.

The simplest approach would be to simply use the beta obtained using the empirical
regression above. This suggests that for the particular SVS strategy discussed above, one is
required to hold a $18 short SPX position for every $100 capital in the strategy. Note that
this delta is of course dependent on the leverage or the vega being sold (in this case 0.2 per
$100 of capital).

However, this approach is clearly too crude. Recall that the beta calculation is based on
monthly returns (the returns over the entire life of the variance swap) and presumably one
should ideally adjust the hedge as the variance swap approaches maturity. In addition, the
beta is clearly based on long term average reactivity and does not use any forward looking
information imbedded in option prices. In the next few sections we develop a methodology
to infer the variance swap delta from option prices.

Understanding Variance Swap Delta

Variance Swap Replication
In order to understand the variance swap delta, it is first useful to briefly review the variance
swap replication methodology. The key insight which emerged from a series of papers in
the mid-1990s was that a long variance swap can be replicated by a static position in a
portfolio of vanilla option contracts and a dynamic position in the underlying index.

Thus consider a long position in a variance swap with maturity initiated at time 0 with an
initial strike of 0 and a $1 variance notional on an underlying which evolves according
to:

Where the volatility can be an arbitrary function of and also can be stochastic. However,
we assume that there are no jumps possible. If we assume that a constant interest rate
and continuous dividend yield , we can recast this in terms of the forward
as:

Since we need to calculate the variance swap delta not just at the start but at all times
before expiration, we start with the payoff as of an intermediate time . Then the payoff

28 April 2014 5
Barclays | Special Report

(which occurs at time ), can be written as a sum of the accrued (known) and mark-to-
market (unknown) payoffs as:

0 ,

is the past realized variance and is the future (unknown) realized variance.
is the remaining life of the variance swap. The crucial insight is that the future realized
variance can be written as:

1 1
log
2 2

is the value of the forward on the index expiring at time as of time (Thus ). The
first term can be replicated by a dynamic position in the forward on the underlying index.
The second term is simply a function of the final value of the index and this log payoff can
be replicated using a series of calls and puts. If we assume that the boundary between puts
and calls we use occurs at strike , it can be shown that:

log log ,0 ,0

Hence the payoff can be written as:

2 2 2
0 log

2
,0 ,0

Note that the above equations at this stage do not involve any expectations and so the
above replication is valid for an arbitrary path. A particularly powerful feature of this
replication methodology is that it is model independent. The dynamic hedging required is
obviously model independent and, given its static nature, so is the option position.

We emphasize again that only the second term leads to a dynamic position in the
underlying; the other terms are static exposures and do not need any rebalancing. We need
to distinguish the replication delta, which is required to replicate the variance swap, from
the variance swap delta that we are interested in. An investor seeking to invest in a SVS
strategy is not interested in replication. The goal is to simply remove the market exposure of
the variance swap. We next look at two alternative (but equivalent) ways of calculating the
variance swap delta.

Variance swap delta by directly using the replicating portfolio

Since the variance swap can be replicated by the above portfolio, the variance swap delta is
simply the sum of the replication delta, the delta of the forward and the delta of the
option strip.

The first term in square brackets is independent of the final value of the underlying and
thus not relevant for replication (although it is relevant when we calculate the fair value
of the variance strike).

The second term is the return on a forward contract which is also equal to the excess
return of holding $1 of the underlying on every day. Although the
returns are calculated daily, the payout is only made at time and thus the daily returns
will incur a (positive or negative) interest rate cost when carried until time . Thus the

28 April 2014 6
Barclays | Special Report

number of contracts of the underlying index that needs to be held for a $1 variance
notional is with the term offsetting the interest rate cost.

The third term is a static position in the forward which can be replicated with a short
position of contracts in the underlying. We need the since the stock will also
pay dividends.

Finally, the fourth term is a static position in vanilla options with weight of for a
strike option.

Adding the last three terms we get:

Where and are the put and call deltas for each option.

Variance swap delta by calculating the derivative of expected P&L

We next present an alternative more automatic approach. The idea is to calculate the risk-
neutral expected value of the mark to market P&L, and differentiate it with respect
to as we would do for any derivative contract. Taking expectations for the expression for
future realized volatility, we get:

1 1
log
2 2

Where we introduce the variance strike, as of time for the variance swap expiring
at time and we have used the following two identities:

, 0

Using , we can then finally write the risk-neutral expected value of the mark-
to-market P&L as:

2 2
log

Setting 0 0, gives us the standard VIX formula for 0 . In the usual VIX
formula the terms in the second square brackets are expanded to second order in . The
expansion works since we can always choose to be very close to . However, at a later
time this quantity is not necessarily small and so we cannot make this approximation.

Differentiating the above expression with respect to , the variance swap delta, , at time
can be written as:

This is identical to the expression derived in the previous section.

Note that in order to calculate the option deltas, we do need an explicit option pricing
model. In other words, the variance swap delta is explicitly model-dependent. This is as it
should be because the non-zero variance swap delta is an empirical fact and so one needs a

28 April 2014 7
Barclays | Special Report

model which properly captures the actual dynamics of the underlying. We address this in
the next section.

Accuracy of the replicating portfolio: An empirical examination

The above results depend on several assumptions:

The derivation of the replication portfolio breaks down if the underlying has jumps,
which SPX clearly does.

The exact replication portfolio requires us to trade a continuum of strikes from zero to
infinity. In reality we only have a limited number of discrete strikes.

The derivation assumes that the realized volatility in the variance swap is calculated
using continuous time. The typical variance swap contract calculates the realised
volatility using daily moves.

The replication delta is assumed to be rebalanced continuously, which is of course

impractical.

Thus it is interesting to empirically check how well the replicating portfolio using available
listed options is able to capture the actual payoff of the variance swap. Figure 6 shows the
result of precisely this exercise. To be specific, on each SPX option expiration, we construct
the static replicating portfolio using all the available one month listed SPX options with non-
zero bid price. The variance swap strike is also calculated using the same option strip using
the standard VIX formula. At the close of every day over the next month we do the required
model-independent rebalancing. The actual payoff of the variance swap is then the
difference between the implied and realized variance calculated using the daily returns. The
payoff of the replicating portfolio is the final value of the static option strip on expiration
and the P&L due to the daily rebalancing.

FIGURE 6
Historically, variance swap payoff can be accurately replicated using the replicating
portfolio
Monthly Returns of SVS Strategy Difference in Returns
4% 0.5%
2% 0.4%
0% 0.3%
-2% 0.2%
-4%
0.1%
-6%
0.0%
-8%
-0.1%
-10%
-12% -0.2%
-14% -0.3%
-16% -0.4%
-18% -0.5%
Feb-96 Feb-99 Feb-02 Feb-05 Feb-08 Feb-11 Feb-14
Difference in Returns (RHS) Replication Portfolio Actual
Source: Barclays Research, OptionMetrics, Bloomberg
Note: The SVS strategy sells 0.2 vega of 1M variance swap for $100 of capital.

As shown in Figure 6, the replicating portfolio has done a remarkable job in reproducing the
actual variance swap payoff even during periods of extreme volatility. Thus the effects of
discrete hedging and jumps have historically not been very material.

28 April 2014 8
Barclays | Special Report

Approximations for the Variance Swap Delta

In this section, we explicitly calculate the variance swap delta using a series of better
approximations. We first verify that the variance swap delta in a Black Scholes world (no
smile) is zero. In the presence of a smile, the simplest approximation is to apply the Black-
Scholes formula to calculate the deltas for each option. Although this sticky strike delta is
non-zero it does not fully remove the market beta since it ignores the fact that strike
volatilities themselves have market dependence. We next develop a methodology to
account for strike volatility dynamics using the Skew Stickiness Ratio (SSR) approach we
have used in past publications.

Variance Swap Delta in a Black-Scholes world

As a first step we calculate the variance swap delta assuming that the index follows the
ideal Black-Scholes dynamics. In this case it is simpler to start with:
1
log
2

It is simple to calculate the integral for a geometric distribution and we get:

1
log
2

Since this is independent of , the delta is zero. Thus in the Black-Scholes world, the
replication delta is exactly opposite to the log-contract delta. This makes sense since in this
world the realized volatility is explicitly constant. Thus if options are priced using a different
volatility, the variance swap will capture the VRP independent of the index return.

The real world is of course not that of the Black-Scholes type. The smile to a large extent
reflects the option markets attempt to capture the non-normal dynamics of the underlying
index. As a result in the presence of volatility smile, the log-contract delta will not be equal
to the replicating delta and thus variance swap delta will be non-zero.

Ideally, we should use a full stochastic volatility model capable of reproducing the volatility
surface and then calculate the option strip delta using this model. Since that would be a
non-trivial exercise, we next present a series of simple approximations which we believe are
adequate for our purpose.

Variance Swap Delta in the presence of a smile: Sticky Strike Approach

The simplest approximation for the variance swap delta is to calculate the delta of the
option strip using the Black-Scholes formula for the delta. Thus we use the different implied
volatilities for each strike but use the standard Black-Scholes formula for the delta.

2
, ,

Figure 7 plots the sticky strike variance swap delta calculated using the above formula. For
reference, we also plot the variance swap delta of $18 calculated using the simple
regression result. We see that this option based variance swap delta is able to capture both
the effect of time to maturity remaining and the variation in market conditions.

28 April 2014 9
Barclays | Special Report

FIGURE 7
Sticky Strike Delta is able to vary according to market conditions
Sticky Strike Delta (in $) required to hedge SVS
5

-5

-10

-15

-20

-25

-30

-35
Jan-96 Jan-99 Jan-02 Jan-05 Jan-08 Jan-11 Jan-14

Source: Barclays Research, OptionMetrics, Bloomberg

Note: The SVS strategy sells 0.2 vega of 1M variance swap for $100 of capital. See text for more details.

To remain faithful to the replication strategy, strictly speaking we should set up the option
strip at trade initiation and simply calculate the delta of this portfolio. However, there are
two disadvantages to this approach. First, as the underlying index moves, the initially out of
the money (OTM) options will become in the money (ITM), and become relatively more
illiquid. As a result, the deltas calculated using the listed prices are likely to be more noisy.
One way to mitigate this is to calculate the deltas using the OTM options and convert those
to ITM deltas using put call parity.

A more serious problem is that the initial portfolio is limited to options with non-zero bid
price. This means that if the underlying moves significantly, we might end up in a situation
where we have very few OTM options. We could address this problem by holding a position
in all the listed strikes irrespective of the bid price. An alternative and completely equivalent
approach is to simply to reset the , every day. We emphasize that although is typically
chosen to be as close to the forward as possible, the choice is completely arbitrary from a
purely mathematical point of view. If we choose a different , the change in the option
delta will be completely offset by a change in the static forward (1/ ) term.

In Figure 8, we show the time series of un-hedged short 1M variance swap and short 1M
variance swap hedged with delta using the sticky strike approach. In Figure 9, we show the
rolling correlation of the two strategies with SPX returns. From the figure, we can see that
the correlation of delta-hedged strategy is less than the correlation of the un-hedged
strategy but it is still not equal to zero. In general, this correlation is consistently above zero.
This obviously is due to the fact that strike volatilities themselves increase when spot
declines and thus sticky strike delta under-hedges the variance swap.

28 April 2014 10
Barclays | Special Report

FIGURE 8 FIGURE 9
Hedged SVS using sticky strike delta has lower returns and But the correlation of the hedged SVS with SPX is still
also lower draw-downs mostly positive
350 Rolling 6M Correlation to SPX
100%
300 80%
60%
250
40%
20%
200
0%
-20%
150
-40%

100 -60%
Jan-96 Jan-00 Jan-04 Jan-08 Jan-12 Jan-96 Jan-00 Jan-04 Jan-08 Jan-12
Unhedged SVS Hedged SVS Unhedged SVS Hedged SVS
Source: Barclays Research, OptionMetrics, Bloomberg Source: Barclays Research, OptionMetrics, Bloomberg
Note: The SVS strategy sells 0.2 vega of 1M variance swap for $100 of capital. Note: The SVS strategy sells 0.2 vega of 1M variance swap for $100 of capital.
Hedged SVS here is using sticky strike delta for hedging. Hedged SVS here is using sticky strike delta for hedging.

Approximation for the Sticky Strike Delta

Since the variance swap delta in the sticky strike approach involves a summation over all the
Black-Scholes deltas, it is hard to get any intuition of its properties. However, as we discuss
next, we can produce a simple formula based on the approximation for the fair variance
swap strike. This approach will also help us to go beyond the sticky-strike approximation.

The key insight is that the above approach implicitly assumes that the implied volatilities
follow a sticky-strike model where the strike volatilities are assumed to remain constant. As
we have discussed in previous publication (Special Report: Understanding VVIX) it is
possible to write down simple approximations for the variance swap strike using simple
parameterizations for the volatility smile curve. In particular, assume that the smile curve
can be well approximated as a simple quadratic in the log-strike ( log / ). I.e.:

Then it can be shown that the variance strike can be written as:

1 3 2

Once the problem is reformulated in this manner, the calculation of the variance swap delta
is thus reduced to question of how the volatility curve parameters change as the underlying
changes.

As discussed above using the Black-Scholes formula for the delta is equivalent to assuming
that that strike volatilities do not change. In terms of the above formulation it means that
when, ,

/ ; 2 / ;

This in turn implies that the change in fair variance or the variance swap strike can be
written as:

2 3 18 6 2

28 April 2014 11
Barclays | Special Report

This directly gives an estimate for the variance swap delta as a function of the curve
parameters:

2 3 18 6 2

To judge the accuracy of this formula we start with a specific parameterization for the
volatility surface using typical empirical values for SPX and then calculate the variance swap
delta by directly pricing the strip of options and by using the above approximate formula. As
shown in Figure 10 the above formula closely replicates the actual variance swap delta
calculated using the actual option deltas. For simplicity we only show the delta calculated at
trade initiation (one month to expiry).

FIGURE 10
Delta from option strip is close to the delta calculated from approximate variance strike
formula
Delta (in $) required to hedge SVS
0

-5

-10

-15

-20

-25
Jan-96 Jan-00 Jan-04 Jan-08 Jan-12
Strip Approximate Formula
Source: Barclays Research, OptionMetrics, Bloomberg
Note: The SVS strategy sells 0.2 vega of 1M variance swap for $100 of capital.

This approach confirms that the delta of the variance swap in the sticky strike world can be
thought of essentially the change in the ATM volatility as spot moves up or down the skew
curve. However, we also get some additional insights:

The delta will generally increase as the ATM volatility and skew increase. This has
important consequences for risk management of SVS strategy. Essentially, as a risk-off
phase develops, volatility and skew will both increase, leading to a bigger short position
in the index which is of course desirable.

There is an explicit dependence on time to maturity because of the term, which simply
reflects the fact that for fixed variance notional the vega will increase with time to
maturity. However, for fixed vega, the time dependence of ATM volatility (increasing)
and skew (decreasing) will also lead to additional dependence of the delta on time to
maturity.

Going beyond the Sticky Strike Delta: The SSR approach

We next discuss how to incorporate the strike volatility reactivity to calculate the delta of
the variance swap. In our previous publication Index Volatility Weekly: Are Investors Short
Volatility?, we had used a simple way to quantify the movement in strike volatilities by
regressing changes in ATM volatility versus the returns of the underlying index normalized
by the skew ( ):

28 April 2014 12
Barclays | Special Report

Thus, in a world where the options follow a sticky strike model, the SSR (Skew Stickiness
Ratio) would be equal to 1. In case strike volatilities themselves increase as the spot
declines, the SSR would be greater than 1.

It can be shown that for a local volatility model, the SSR is always 2. The situation for more
general stochastic models is more complicated. In the absence of jumps for very small
expirations, the SSR is still 2 but approaches 1 for very large expirations. Presence of jumps
will decrease the SSR ratio. Intuitively, this is because part of the skew can now be
attributed to the jumps.

In Figure 11, we show the empirical SSR for 1M, 2M and 3M SPX volatility using 2-year
rolling window. We see several interesting trends:

In general, the SSR does deviate significantly from 1 and is typically higher than 1.

The SSR appears to have gone through some regime shifts. It used to be much closer to
1 prior to 2007 and appears to have reached a higher level post the 2007-08 crises.

In general, there does not appear to be much of dependence on option maturity. Thus
the SSR ratios calculated using 1M, 2M and 3M options appear to be quite close.

FIGURE 11
SSR is higher than 1 and appears to have increased after 2007
Rolling 2Y SSR
2.0
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
Dec-97 Dec-01 Dec-05 Dec-09 Dec-13
1M IV 2M IV 3M IV
Source: Barclays Research, OptionMetrics, Bloomberg
Note: SSR is calculated using trailing 2-year regression of Daily Change in ATM IV ~ Skew*(Daily Returns of SPX).

The simplest way to relax the sticky strike approximation would be to simply use the
median value of the SSR over the entire history and simply scale up the sticky strike variance
swap delta. However, since the SSR value does appear to go through regime shifts, it makes
sense to use the rolling value of the SSR at any given time. This also avoids look-ahead bias
in our back-tests.

Performance of market-neutral variance swap strategies

Our basic approach is to trade the variance swap and then overlay a short position in the
underlying index with the calculated variance swap delta. We first calculate the option strip
delta using the sticky-strike approach and then adjust it by the SSR ratio by using the
trailing two-year empirical SSR ratio. Since the additional position provides an automatic
risk-management mechanism, it is also interesting to examine what happens if you over-
hedge by using a SSR of 2. In Figure 12 and Figure 13, we show the time series and
performance metrics of these different strategies. We show the results with and without

28 April 2014 13
Barclays | Special Report

transaction costs. We also include the performance of an ATM delta-hedged straddle as a

point of comparison.

FIGURE 12
While the hedging does reduce the overall returns of SVS, the hedged SVS still appears to
be better than short-delta-hedged straddles
340

290

240

190

140

90
Jan-96 Jan-00 Jan-04 Jan-08 Jan-12
Unhedged SVS Hedged SVS (SSR =1)
Hedged SVS (Rolling SSR) Hedged SVS (SSR = 2.0)
Delta Hedged ATM Straddle
Source: Barclays Research, OptionMetrics, Bloomberg
Note: The SVS strategy sells 0.2 vega of 1M variance swap for $100 of capital. Transaction costs are ignored.

FIGURE 13
The Sharpe ratio of hedged SVS is not significantly different from unhedged SVS,
attesting to the robust nature of VRP

Without Transaction Cost With Transaction Cost

Max 90% Corr.

Sharpe
Strategy Avg. Ret. Std. Dev. Draw- Monthly With Avg. Ret. Sharpe Ratio
Ratio
down cVaR SPX

Unhedged SVS 6.5% 5.2% 1.25 -24.2% 2.6% 58% 5.4% 1.03

Hedged SVS
5.5% 4.0% 1.38 -21.1% 2.3% 26% 4.4% 1.07
(SSR =1)
Hedged SVS
5.2% 3.8% 1.36 -19.4% 2.2% 4% 3.9% 1.01
(Rolling SSR)

Hedged SVS
4.6% 3.9% 1.19 -18.4% 2.3% -22% 3.3% 0.83
(SSR = 2.0)

Delta Hedged
3.3% 4.1% 0.81 -10.9% 1.7% 25% 2.1% 0.52
ATM Straddle
Source: Barclays Research, OptionMetrics, Bloomberg. Performance as of April 17, 2014
Note: The SVS strategy sells 0.2 vega of 1M variance swap for $100 of capital. For transaction cost, we have assumed
0.5 vol points of bid-ask spread for variance swaps, 2bps for delta-hedging and 2bps of the notional for SPX options.

From the above results, we can make several interesting observations:

Correlation of hedged SVS with SPX: First, we notice that the new strategy using
rolling empirical SSR has very low correlation with SPX; thus this new method is
effective in capturing the delta of the short variance swap strategy. In order to
understand the effect of hedging, in Figure 14 and Figure 15 we also show the rolling
correlation of the hedged variance swap and also the average returns profile of hedged

28 April 2014 14
Barclays | Special Report

SVS with respect to monthly SPX returns. From Figure 14, we also observe that this low
correlation of the hedged SVS with SPX has been effective in different regimes and as
such we do not see any biases in it. Figure 15 shows that the average monthly returns of
SVS are now insensitive to moderate monthly returns in SPX.

Impact on draw-downs: Figure 15 also demonstrates that the exposure to large

moves remains significant. The draw-downs of hedged SVS are not significantly lower
than the draw-downs of un-hedged SVS. This is because of the strong convexity SVS
has relative to the SPX returns, and as such hedging can reduce the decline in returns
only when SPX declines are moderate (5%-10% per month).

Alpha of hedged SVS: We see that even the market neutral variance swap has high
Sharpe ratio and thus, it has significant positive alpha. Thus, the VRP appears to be a
robust phenomenon and does not completely disappear after the market beta is
removed.

Implications of over-hedging: An over-hedging SVS strategy actually results in a

negative correlation of SVS with SPX. This reduces the Sharpe ratio of the strategy as
the standard deviation starts to increase and average returns decline because of drift in
SPX.

Short ATM delta hedged straddles correlation to SPX: Delta-hedged ATM

straddles using Black-Scholes implied volatility have a similar correlation to SPX as
delta-hedged SVS using SSR of 1. This is because the delta calculated from Black-
Scholes does not take into account the increase in fixed strike implied volatilities when
SPX declines. A more refined model, which takes into account the change in strike
volatilities using SSR greater than 1, would have lower correlation to SPX.

Difference between short delta-hedged ATM straddles and SVS: The market
neutral SVS strategy also allows us to perform a more robust comparison between
variance swaps and delta-hedged straddles. As can be seen, delta hedged straddles have
historically underperformed variance swaps. The typical reasons advanced for this are
that the former suffer from path dependency and also do not directly monetize the rich
skew imbedded in OTM puts. However, as discussed above, another important
difference is the implicit market exposure in variance swaps. Thus, comparing delta-
hedged straddles with delta-hedged variance swaps provides a much better way to
quantify the impact of path-dependency and skew. Hedged SVS has a much higher
Sharpe ratio but a hedged ATM straddle strategy has much fewer draw-downs. This is
because selling rich implied volatility OTM put options is on average better, except
during significant sell-offs. In the case of short ATM delta-hedged straddles, during
severe sell-offs (like 2008), the underlying prices move significantly away from the strike
and thus the ATM delta-hedged straddle strategy has very little gamma and thus losses
are less.

Performance in bullish periods: In bullish periods (like 2013), as expected the

performance of un-hedged SVS is significantly better than hedged SVS (in 2013: hedged
and un-hedged SVS had returns of 1% and 6%, respectively). But even hedged SVS
performs much better than short delta-hedged straddles as hedged SVS benefits from
selling OTM puts (short ATM delta-hedged straddles lost 1.14% in 2013).

28 April 2014 15
Barclays | Special Report

FIGURE 14 FIGURE 15
Hedged SVS has low correlation irrespective of different Hedged SVS has low sensitivity to moderate SPX returns but
regimes declines significantly for large SPX returns

Rolling 6M Correlation to SPX SVS 1M Ret.

100% 5%
80%
60% 0%

40%
-5%
20%
0%
-10%
-20%
-40% -15%
-60%
-80% -20%
Jan-96 Jan-00 Jan-04 Jan-08 Jan-12 -30% -20% -10% 0% 10%
SPX 1M Ret.
Unhedged SVS Hedged SVS (Rolling SSR)
Unhedged Hedged (Rolling SSR)
Source: Barclays Research, OptionMetrics, Bloomberg Source: Barclays Research, OptionMetrics, Bloomberg
Note: The SVS strategy sells 0.2 vega of 1M variance swap for $100 of capital. Note: The SVS strategy sells 0.2 vega of 1M variance swap for $100 of capital.

Market Neutral Short VIX Futures Strategy

As we have documented extensively in our regular publications (Systematic Volatility
Monthly and The VIX Compass), a short position in VIX futures has historically resulted in
attractive risk reward characteristics. This results from the premium that exists in the
volatility term structure. In other words, the term structure of volatility tends to be too
steep. However, the term structure premium has historically not been as robust as the
volatility risk premium. In this section, we apply the above methodology to calculate the
delta of VIX futures and use that to create market neutral short VIX future strategies.

For simplicity, we proxy VIX futures by forward variance swaps. This is an approximation
since the payoff for VIX futures is linear in volatility versus the quadratic dependence for
forward variance swaps. This difference in the payoffs is what leads VIX futures to trade at a
discount to forward variance swaps leading to a convexity adjustment which is positively
dependent on volatility of volatility. Below we simply use the delta of the forward variance
swap to hedge the VIX futures and ignore the convexity effect.

A forward variance swap is simply a long-short combination of variance swaps of two

maturities. Thus, a forward variance swap strike can be written as:

Here , is the strike of forward variance swap with start date and end date and
and are the strikes for spot variance swaps maturing at and respectively.
Differentiating the above equation with respect to spot price, we see the forward variance
swap delta can easily be expressed in terms of spot variance swap deltas.

We can thus calculate the delta of the VIX future using the (SSR adjusted) deltas of the spot
variance swaps. The forward variance swaps calculated using the listed SPX option
expirations do not line up exactly with the tenors of the VIX futures which always expire
exactly 30 days before SPX expiration. We ignore this misalignment of 2-3 business days in
our calculation o f the delta.
28 April 2014 16
Barclays | Special Report

In Figure 16 and Figure 17, we show the performance of an un-hedged and hedged short
1M VIX futures strategy. The strategy systematically shorts the front month VIX future and
rolls it 10 days before expiration. We have sized the short VIX futures strategy to be equal to
0.4 vega per $100 in equity to make the draw-downs of short VIX futures strategy
comparable to SVS.

FIGURE 16
Hedged Short VIX futures strategy has very low standard deviation

130

120

110

100

90

80

70
Jan-06 Jan-07 Jan-08 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14
Unhedged Hedged
Source: Barclays Research, Bloomberg, OptionMetrics
Note: The Short VIX futures strategy sells 0.4 vega of front month VIX futures for $100 of capital. The VIX futures are
rolled when they have less than 10 days to expiry. Hedging is carried out using rolling SSR. No transaction costs are
taken into account.

FIGURE 17
Similar to SVS, the Sharpe ratio of hedged Short VIX futures strategy is not significantly
different from the unhedged strategy
90%
Max Draw- Corr. With
Strategy Avg. Ret. Std. Dev. Sharpe Ratio Monthly
down SPX
cVaR
Unhedged Short VIX
3.3% 7.3% 0.46 -25.8% 4.9% 83%
Futures
Hedged Short VIX Futures 1.6% 3.9% 0.41 -15.0% 3.2% 11%

Source: Barclays Research, Bloomberg, OptionMetrics. Performance till April 17, 2014
Note: The Short VIX futures strategy sells 0.4 vega of front month VIX futures for $100 of capital. The VIX futures are
rolled when they have less than 10 days to expiry. Hedging is carried out using rolling SSR. No transaction costs are
taken into account.

We see that the hedged short VIX futures strategy shows many similarities to hedged SVS.

As shown in more detail in Figure 18 , the correlation of the un-hedged VIX futures
strategy with market returns is very high (~80%). However, the correlation of hedged
short VIX futures strategy drops substantially and remains low across different market
regimes.

There is positive alpha in short VIX futures strategy even after accounting for market
exposure (the Sharpe ratio of un-hedged and hedged short VIX futures strategy is 0.46
and 0.41, respectively). Thus the term structure premium does not completely disappear
after adjusting for the market exposure.

In contrast to SVS, the draw-down reduction is higher for the short VIX futures strategy.
This is because the convexity of short VIX futures strategy to SPX returns is less in
comparison to SVS.

28 April 2014 17
Barclays | Special Report

Similar to hedged SVS, the hedged short VIX futures strategy had low returns (-0.7%) in
2013 when equity markets rallied significantly.

FIGURE 18 FIGURE 19
Hedged short VIX futures strategy has low correlation in Hedged short VIX futures strategy has low sensitivity to
different periods moderate SPX returns
Rolling 6M Correlation to SPX Short VIX future 1M Ret
100% 5%
80%
60% 0%

40%
-5%
20%
0%
-10%
-20%
-40% -15%
-60%
-80% -20%
-100% -30% -20% -10% 0% 10%
Jan-06 Jan-08 Jan-10 Jan-12 Jan-14 SPX 1M Ret
Unhedged Hedged Unhedged Hedged
Source: Barclays Research, Bloomberg, OptionMetrics. Source: Barclays Research, Bloomberg, OptionMetrics
Note: The Short VIX futures strategy sells 0.4 vega of front month VIX futures for Note: The Short VIX futures strategy sells 0.4 vega of front month VIX futures for
$100 of capital. $100 of capital.

Summary
While variance swaps are a pure means to trade volatility, they have significant market
exposure (variance swap delta). This variance swap delta does not exist in an ideal Black-
Scholes world (no volatility smile), but is non-zero when the volatility surface exhibits a
smile. One way to calculate the delta of the variance swap in the presence of a smile is by
adding the replicating delta of the variance swap with the delta of the option strip calculated
from Black-Scholes. This delta calculation assumes the strike volatility does not react when
spot moves (sticky strike), which is empirically not true. We show that the delta calculated
from the sticky strike model can be adjusted for strike volatility dynamics by using empirical
SSR.

The SVS hedged with the delta adjusted for SSR has very low correlation to SPX returns. The
hedged SVS has a Sharpe ratio similar to the un-hedged SVS, which attests to the robust
nature of the volatility risk premium. While the hedging reduces draw-downs for moderate
declines in the SPX, it does not reduce draw-downs significantly when the SPX declines are
large, as the SVS has high convexity relative to SPX returns.

Using the above approach, delta of the VIX futures can also be calculated by calculating the
delta of the corresponding forward variance swap. Similar to SVS, the Sharpe ratio of a short
VIX futures strategy hedged using the delta calculated from this method is similar to an un-
hedged short VIX futures strategy.

28 April 2014 18
Barclays | Special Report

ANALYST(S) CERTIFICATION(S):
I, Maneesh S. Deshpande, 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 research report and (2) no part of my compensation was, is or will be directly or indirectly
related to the specific recommendations or views expressed in this research report.

IMPORTANT DISCLOSURES CONTINUED

Barclays Research is a part of the Corporate and Investment Banking division of Barclays Bank PLC and its affiliates (collectively and each
individually, "Barclays"). For current important disclosures regarding companies that are the subject of this research report, please send a written
request to: Barclays Research Compliance, 745 Seventh Avenue, 14th Floor, New York, NY 10019 or refer to http://publicresearch.barclays.com
or call 212-526-1072.
The analysts responsible for preparing this research report have received compensation based upon various factors including the firm's total
revenues, a portion of which is generated by investment banking activities.
Analysts regularly conduct site visits to view the material operations of covered companies, but Barclays policy prohibits them from accepting
payment or reimbursement by any covered company of their travel expenses for such visits.
In order to access Barclays Statement regarding Research Dissemination Policies and Procedures, please refer to
https://live.barcap.com/publiccp/RSR/nyfipubs/disclaimer/disclaimer-research-dissemination.html. In order to access Barclays Research
Conflict Management Policy Statement, please refer to: http://group.barclays.com/corporates-and-institutions/research/research-policy.
The Corporate and Investment Banking division of Barclays produces a variety of research products including, but not limited to, fundamental
analysis, equity-linked analysis, quantitative analysis, and trade ideas. Recommendations contained in one type of research product may differ
from recommendations contained in other types of research products, whether as a result of differing time horizons, methodologies, or
otherwise.

Risk Disclosure(s)
Options are not suitable for all investors. Please note that the trade ideas within this research report do not necessarily relate to, and may directly
conflict with, the fundamental ratings applied to Barclays Equity Research. The risks of options trading should be weighed against the potential
rewards.
Risks:
Call or put purchasing: The risk of purchasing a call/put is that investors will lose the entire premium paid.
Uncovered call writing: The risk of selling an uncovered call is unlimited and may result in losses significantly greater than the premium received.
Uncovered put writing: The risk of selling an uncovered put is significant and may result in losses significantly greater than the premium received.
Call or put vertical spread purchasing (same expiration month for both options): The basic risk of effecting a long spread transaction is limited to
the premium paid when the position is established.
Call or put vertical spread writing/writing calls or puts (usually referred to as uncovered writing, combinations or straddles; same expiration
month for both options): The basic risk of effecting a short spread transaction is limited to the difference between the strike prices less the
amount received in premiums.
Call or put calendar spread purchasing (different expiration months; short must expire prior to the long): The basic risk of effecting a long
calendar spread transaction is limited to the premium paid when the position is established.
Because of the importance of tax considerations to many options transactions, the investor considering options should consult with his/her tax
advisor as to how taxes affect the outcome of contemplated options transactions.
Supporting documents that form the basis of our recommendations are available on request.
The Options Clearing Corporation's report, "Characteristics and Risks of Standardized Options", is available at
http://www.theocc.com/about/publications/character-risks.jsp
Barclays offices involved in the production of equity research:
London
Barclays Bank PLC (Barclays, London)
New York
Barclays Capital Inc. (BCI, New York)
Tokyo
Barclays Securities Japan Limited (BSJL, Tokyo)
So Paulo
Banco Barclays S.A. (BBSA, So Paulo)
Hong Kong
Barclays Bank PLC, Hong Kong branch (Barclays Bank, Hong Kong)
Toronto

28 April 2014 19
Barclays | Special Report

IMPORTANT DISCLOSURES CONTINUED

Barclays Capital Canada Inc. (BCCI, Toronto)
Johannesburg
Absa Bank Limited (Absa, Johannesburg)
Mexico City
Barclays Bank Mexico, S.A. (BBMX, Mexico City)
Taiwan
Barclays Capital Securities Taiwan Limited (BCSTW, Taiwan)
Seoul
Barclays Capital Securities Limited (BCSL, Seoul)
Mumbai
Barclays Securities (India) Private Limited (BSIPL, Mumbai)
Singapore
Barclays Bank PLC, Singapore branch (Barclays Bank, Singapore)

28 April 2014 20
DISCLAIMER:
This publication has been prepared by the Corporate and Investment Banking division of Barclays Bank PLC and/or one or more of its affiliates (collectively
and each individually, "Barclays"). It has been issued by one or more Barclays legal entities within its Corporate and Investment Banking division as provided
below. It is provided to our clients for information purposes only, and Barclays makes no express or implied warranties, and expressly disclaims all warranties
of merchantability or fitness for a particular purpose or use with respect to any data included in this publication. Barclays will not treat unauthorized
recipients of this report as its clients. Prices shown are indicative and Barclays is not offering to buy or sell or soliciting offers to buy or sell any financial
instrument.
Without limiting any of the foregoing and to the extent permitted by law, in no event shall Barclays, nor any affiliate, nor any of their respective officers,
directors, partners, or employees have any liability for (a) any special, punitive, indirect, or consequential damages; or (b) any lost profits, lost revenue, loss of
anticipated savings or loss of opportunity or other financial loss, even if notified of the possibility of such damages, arising from any use of this publication or
its contents.
Other than disclosures relating to Barclays, the information contained in this publication has been obtained from sources that Barclays Research believes to
be reliable, but Barclays does not represent or warrant that it is accurate or complete. Barclays is not responsible for, and makes no warranties whatsoever as
to, the content of any third-party web site accessed via a hyperlink in this publication and such information is not incorporated by reference.
The views in this publication are those of the author(s) and are subject to change, and Barclays has no obligation to update its opinions or the information in
this publication. The analyst recommendations in this publication reflect solely and exclusively those of the author(s), and such opinions were prepared
independently of any other interests, including those of Barclays and/or its affiliates. This publication does not constitute personal investment advice or take
into account the individual financial circumstances or objectives of the clients who receive it. The securities discussed herein may not be suitable for all
investors. Barclays recommends that investors independently evaluate each issuer, security or instrument discussed herein and consult any independent
advisors they believe necessary. The value of and income from any investment may fluctuate from day to day as a result of changes in relevant economic
markets (including changes in market liquidity). The information herein is not intended to predict actual results, which may differ substantially from those
reflected. Past performance is not necessarily indicative of future results.
This material has been issued and approved for distribution in the UK and European Economic Area (EEA) by Barclays Bank PLC. It is being made available
primarily to persons who are investment professionals as that term is defined in Article 19 of the Financial Services and Markets Act 2000 (Financial
Promotion) Order 2005. It is directed at, and therefore should only be relied upon by, persons who have professional experience in matters relating to
investments. The investments to which it relates are available only to such persons and will be entered into only with such persons. Barclays Bank PLC is
authorised by the Prudential Regulation Authority and regulated by the Financial Conduct Authority and the Prudential Regulation Authority and is a
member of the London Stock Exchange.
The Corporate and Investment Banking division of Barclays undertakes U.S. securities business in the name of its wholly owned subsidiary Barclays Capital
Inc., a FINRA and SIPC member. Barclays Capital Inc., a U.S. registered broker/dealer, is distributing this material in the United States and, in connection
therewith accepts responsibility for its contents. Any U.S. person wishing to effect a transaction in any security discussed herein should do so only by
contacting a representative of Barclays Capital Inc. in the U.S. at 745 Seventh Avenue, New York, New York 10019.
Non-U.S. persons should contact and execute transactions through a Barclays Bank PLC branch or affiliate in their home jurisdiction unless local regulations
permit otherwise.
Barclays Bank PLC, Paris Branch (registered in France under Paris RCS number 381 066 281) is regulated by the Autorit des marchs financiers and the
Autorit de contrle prudentiel. Registered office 34/36 Avenue de Friedland 75008 Paris.
This material is distributed in Canada by Barclays Capital Canada Inc., a registered investment dealer and member of IIROC (www.iiroc.ca).
Subject to the conditions of this publication as set out above, the Corporate & Investment Banking Division of Absa Bank Limited, an authorised financial
services provider (Registration No.: 1986/004794/06. Registered Credit Provider Reg No NCRCP7), is distributing this material in South Africa. Absa Bank
Limited is regulated by the South African Reserve Bank. This publication is not, nor is it intended to be, advice as defined and/or contemplated in the (South
African) Financial Advisory and Intermediary Services Act, 37 of 2002, or any other financial, investment, trading, tax, legal, accounting, retirement, actuarial
or other professional advice or service whatsoever. Any South African person or entity wishing to effect a transaction in any security discussed herein should
do so only by contacting a representative of the Corporate & Investment Banking Division of Absa Bank Limited in South Africa, 15 Alice Lane, Sandton,
Johannesburg, Gauteng 2196. Absa Bank Limited is a member of the Barclays group.
In Japan, foreign exchange research reports are prepared and distributed by Barclays Bank PLC Tokyo Branch. Other research reports are distributed to
institutional investors in Japan by Barclays Securities Japan Limited. Barclays Securities Japan Limited is a joint-stock company incorporated in Japan with
registered office of 6-10-1 Roppongi, Minato-ku, Tokyo 106-6131, Japan. It is a subsidiary of Barclays Bank PLC and a registered financial instruments firm
regulated by the Financial Services Agency of Japan. Registered Number: Kanto Zaimukyokucho (kinsho) No. 143.
Barclays Bank PLC, Hong Kong Branch is distributing this material in Hong Kong as an authorised institution regulated by the Hong Kong Monetary
Authority. Registered Office: 41/F, Cheung Kong Center, 2 Queen's Road Central, Hong Kong.
Information on securities/instruments that trade in Taiwan or written by a Taiwan-based research analyst is distributed by Barclays Capital Securities
Taiwan Limited to its clients. The material on securities/instruments not traded in Taiwan is not to be construed as 'recommendation' in Taiwan. Barclays
Capital Securities Taiwan Limited does not accept orders from clients to trade in such securities. This material may not be distributed to the public media or
used by the public media without prior written consent of Barclays.
This material is distributed in South Korea by Barclays Capital Securities Limited, Seoul Branch.
All equity research material is distributed in India by Barclays Securities (India) Private Limited (SEBI Registration No: INB/INF 231292732 (NSE), INB/INF
011292738 (BSE) | Corporate Identification Number: U67120MH2006PTC161063 | Registered Office: 208 | Ceejay House | Dr. Annie Besant Road | Shivsagar
Estate | Worli | Mumbai - 400 018 | India, Phone: + 91 22 67196363). Other research reports are distributed in India by Barclays Bank PLC, India Branch.
Barclays Bank PLC Frankfurt Branch distributes this material in Germany under the supervision of Bundesanstalt fr Finanzdienstleistungsaufsicht (BaFin).
This material is distributed in Malaysia by Barclays Capital Markets Malaysia Sdn Bhd.
This material is distributed in Brazil by Banco Barclays S.A.
This material is distributed in Mexico by Barclays Bank Mexico, S.A.
Barclays Bank PLC in the Dubai International Financial Centre (Registered No. 0060) is regulated by the Dubai Financial Services Authority (DFSA). Principal
place of business in the Dubai International Financial Centre: The Gate Village, Building 4, Level 4, PO Box 506504, Dubai, United Arab Emirates. Barclays
Bank PLC-DIFC Branch, may only undertake the financial services activities that fall within the scope of its existing DFSA licence. Related financial products or
services are only available to Professional Clients, as defined by the Dubai Financial Services Authority.
Barclays Bank PLC in the UAE is regulated by the Central Bank of the UAE and is licensed to conduct business activities as a branch of a commercial bank
incorporated outside the UAE in Dubai (Licence No.: 13/1844/2008, Registered Office: Building No. 6, Burj Dubai Business Hub, Sheikh Zayed Road, Dubai
City) and Abu Dhabi (Licence No.: 13/952/2008, Registered Office: Al Jazira Towers, Hamdan Street, PO Box 2734, Abu Dhabi).
Barclays Bank PLC in the Qatar Financial Centre (Registered No. 00018) is authorised by the Qatar Financial Centre Regulatory Authority (QFCRA). Barclays
Bank PLC-QFC Branch may only undertake the regulated activities that fall within the scope of its existing QFCRA licence. Principal place of business in Qatar:
Qatar Financial Centre, Office 1002, 10th Floor, QFC Tower, Diplomatic Area, West Bay, PO Box 15891, Doha, Qatar. Related financial products or services
are only available to Business Customers as defined by the Qatar Financial Centre Regulatory Authority.
This material is distributed in the UAE (including the Dubai International Financial Centre) and Qatar by Barclays Bank PLC.
This material is distributed in Saudi Arabia by Barclays Saudi Arabia ('BSA'). It is not the intention of the publication to be used or deemed as
recommendation, option or advice for any action (s) that may take place in future. Barclays Saudi Arabia is a Closed Joint Stock Company, (CMA License No.
09141-37). Registered office Al Faisaliah Tower, Level 18, Riyadh 11311, Kingdom of Saudi Arabia. Authorised and regulated by the Capital Market
Authority, Commercial Registration Number: 1010283024.
This material is distributed in Russia by OOO Barclays Capital, affiliated company of Barclays Bank PLC, registered and regulated in Russia by the FSFM.
Broker License #177-11850-100000; Dealer License #177-11855-010000. Registered address in Russia: 125047 Moscow, 1st Tverskaya-Yamskaya str. 21.
This material is distributed in Singapore by the Singapore branch of Barclays Bank PLC, a bank licensed in Singapore by the Monetary Authority of Singapore.
For matters in connection with this report, recipients in Singapore may contact the Singapore branch of Barclays Bank PLC, whose registered address is One
Raffles Quay Level 28, South Tower, Singapore 048583.
Barclays Bank PLC, Australia Branch (ARBN 062 449 585, AFSL 246617) is distributing this material in Australia. It is directed at 'wholesale clients' as defined
by Australian Corporations Act 2001.
IRS Circular 230 Prepared Materials Disclaimer: Barclays does not provide tax advice and nothing contained herein should be construed to be tax advice.
Please be advised that any discussion of U.S. tax matters contained herein (including any attachments) (i) is not intended or written to be used, and cannot
be used, by you for the purpose of avoiding U.S. tax-related penalties; and (ii) was written to support the promotion or marketing of the transactions or other
matters addressed herein. Accordingly, you should seek advice based on your particular circumstances from an independent tax advisor.
Copyright Barclays Bank PLC (2014). All rights reserved. No part of this publication may be reproduced or redistributed in any manner without the prior
written permission of Barclays. Barclays Bank PLC is registered in England No. 1026167. Registered office 1 Churchill Place, London, E14 5HP. Additional
information regarding this publication will be furnished upon request.
US08-000001