An Interest Rate Swap Volatility Index and Contract

Antonio Mele
QUASaR
Yoshiki Obayashi
Applied Academics LLC
First draft: November 10, 2009. This version: June 26, 2012.
ABSTRACT Interest rate volatility and equity volatility evolve heterogeneously over time, co-
moving disproportionately during periods of global imbalances and each reacting to events of
dierent nature. While the Chicago Board Options Exchange
R _
(CBOE
R _
) VIX
R _
reects the fair
value of equity volatility, no interest rate counterparts exist to the CBOE VIX
R _
; we ll this
gap for swap rate volatility. We use data on interest rate swaptions and bonds to construct two
indexes of interest rate swap volatility expected to prevail in a risk-neutral market within any
given investment horizon. The two indexes match market practices of quoting swaption-implied
volatilities both in terms of basis point and percentage volatility. The indexes are constructed such
that they reect the model-free fair value of variance swap contracts for forward swap rates that
we design to account for the uncertainty in the annuity factor of the xed leg of forward swaps.
The indexes are model-free in the sense that they do not rely on any particular option pricing
model. While the end-result can be thought of as the swap rate counterpart to the CBOE VIX
R _
for equity volatility, the mathematical formulations, contract designs, replication arguments, and
derivatives underlying the index are materially dierent. The dierences arise as we must account
not only for changes in forward swap rates but also in the yield curve points impacting forward
swap values in a framework where interest rates are not assumed to be constant as they are
in the case of the VIX
R _
methodology. We also consider a framework in which swap rates may
experience jumps and nd that our basis point volatility index can be expressed in a model-free
format even in the presence of discontinuities.
c _ by Antonio Mele and Yoshiki Obayashi
1
An Interest Rate Swap Volatility Index and Contract
Contents
1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2. Interest rate transaction risks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3. Option-based volatility strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
4. Interest rate swap variance contracts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
4.1 Contracts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .10
4.2 Pricing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
4.3 Hedging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
5. Basis point volatility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
5.1 Contracts and evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
5.2 Hedging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
5.3 Links to constant maturity swaps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
6. Marking to market and trading strategies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
6.1 Marks to market of IRV swaps. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
6.2 Trading strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
7. Swap volatility indexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
7.1 Percentage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
7.2 Basis point . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
8. Rates versus equity variance contracts and indexes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .26
9. Resilience to jumps. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
10. Index implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
10.1 A numerical example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
10.2 Historical performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Technical Appendix. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .40
A. Notation, assumptions and preliminary facts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
B. P&L of option-based volatility trading. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
C. Spanning the contracts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
D. Directional volatility trades. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
E. Local volatility surfaces. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
F. The contracts and index in the Vasicek market. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .48
G. Spanning and hedging Gaussian contracts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
H. Jumps. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .58
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
2 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
1. Introduction
Interest rate volatility and equity volatility evolve heterogeneously over time, comoving dispro-
portionately during periods of global imbalances and each reacting to events of dierent nature.
While the Chicago Board Options Exchange
R
(CBOE
R
) VIX
R
reects the fair value of equity
volatility, no interest rate counterparts exist to the CBOE VIX
R
; we ll this gap for swap rate
volatility.
Figure 1 depicts the 10 year spot swap rate over nearly two decades and compares its 20 day
realized volatility with that of the S&P 500
R
Index. While these two volatilities share some
common trends and spikes, one is hardly a proxy for the other. Their average correlation over
the sample size is a mere 51%. Moreover, Figure 2 shows that the correlation between the two
volatilities experiences large swings over time. For instance, before the nancial crisis starting
in 2007, the correlation was negative. In the rst years of this crisis, the same correlation was
high, most likely due to global concerns about disorderly tail events. However, even during
the crisis, we see periods of marked divergences between the two volatilities. After 2010, for
example, the correlation between the two volatilities plummeted, reaching negative values in
2011.
How can one hedge against or formulate views on uncertainties around changes in swap
rates? In swap markets, which are the focus of this paper, volatility is typically traded through
swaption-based strategies such as straddles: for example, going long both a payer and a re-
ceiver swaption may lead to prots should interest rates experience a period of high volatility.
However, delta-hedged or unhedged swaption straddles do not necessarily lead to prots con-
sistent with directional volatility views. The reason for this discrepancy is, at least partially,
similar to that explaining the failure of equity option straddles to deliver P&L consistent with
directional volatility views. Straddles suer from price dependency, the tendency to generate
P&L aected by the direction of price movements rather than their absolute movementsi.e.
volatility. These, and related reasons, have led to variance swap contracts in equity markets
designed to better align volatility views and payos, and a new VIX
R
index maintained by the
Chicago Board Options Exchange since 2003.
Despite the fact that transaction volumes in the swap and swaption markets are orders of
magnitude greater than in equity markets, no such index or variance contracts exist. Moreover,
there are only weak linkages between interest rate swap volatility and equity volatility, as
illustrated by Figures 1 and 2, which further motivates the creation of a VIX-like index for
swap rates. At the same time, the complexity of interest rate transactions is such that their
volatility is more dicult to price. Intuitively, one prices assets by discounting their future
payos, but when it comes to xed income securities, discounting rates are obviously random
and, in turn, interest rate swap volatility depends on this randomness. An additional critical
issue pertaining to swap rate volatility is that uncertainty in swap markets is driven by two
sources of randomness: rst, a direct source relating to changes in swap rates; and second, an
indirect one pertaining to changes in points of the yield curve that aect the value of swap
contracts.
This paper introduces security designs to trade uncertainty related to developments of interest
rate swap volatility while addressing the above issues. The pricing of these products is model-
free, as it only relies on the price of traded assets such as (i) European-style swaptions, serving as
3 c by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
1994 1996 1998 2000 2002 2004 2006 2008 2010 2012
0
5
10
10Y spot Swap Rate (%)
1994 1996 1998 2000 2002 2004 2006 2008 2010 2012
0
50
100
S&P 500 real i zed vol ati l i ty (%)
10Y Swap Rate real ized vol ati l ity (%)
Figure 1. Top panel: The 10 year Swap Rate, annualized, percent. Bottom panel:
estimates of 20 day realized volatilities of (i) the S&P 500
R _
daily returns (solid
line), and (ii) the daily logarithmic changes in the 10 year Swap Rate, annualized,
percent, 100
_
12

21
t=1
ln
2 X
ti+1
X
ti
, where X
t
denotes either the S&P 500 Index
R _
or
the 10 year Swap Rate. The sample includes daily data from January 29, 1993 to
May 22, 2012, for a total of 4865 observations.
1994 1996 1998 2000 2002 2004 2006 2008 2010 2012
-80
-60
-40
-20
0
20
40
60
80
100
Corr. between S&P 500 & SWAP Vol (%)
Figure 2. Moving average estimates of the correlation between the 20 day realized
volatilities of the S&P 500
R _
and the 10 year Swap Rate logarithmic changes. Each
correlation estimate is calculated over the previous one year of data, and the sample
includes daily data from January 29, 1993 to May 22, 2012, for a total of 4865
observations.
4 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
counterparts to the European-style equity options underlying equity volatility (e.g., Demeter,
Derman, Kamal and Zou, 1999), but, also, (ii) the price of zero coupon bonds with maturities
corresponding to the xed payment days of forward swaps underlying swaptions. As such,
our contract designs lead to new indexes that reect market expectations about swap market
volatility, adjusted for risk.
The most basic example of contracts we consider is structured as follows. At time , two
counterparties A and B agree that at time , A shall pay B the realized variance of the forward
swap rate from to , scaled by the price value of a basis point of the xed leg of the forward
swap at the xed payment dates
1
for the payment periods
1

1
(see Figure 3).
t
T T
1 T
n

length of the volatility contract

length of the swap contract (tenor)
Curr ent t ime t. Contract characteristics are
chosen: (i) length of the period over which
the volatility of the forward swap rate will be
accounted, T-t; (ii) tenor of the payments underlying
the time t forward swap rate : T1, , Tn.
T
2
Ti me T. Resolution of uncertainty about the volatility
of the forward swap rate occurred from time t to time T.
Payments related to the volatility contract are executed,
and the contract expires.
Dates T1,,Tn. Dates of execution of the interest
payments relating to the time t forward swap rate.
Figure 3. Timing of the interest rate variance contract and the execution of swap
payments.
As the pricing of this and other variance swap contracts introduced below rely on bonds
and swaptions, the next section provides a concise summary of risks inherent in relevant inter-
est rate transactions. Section 3 highlights pitfalls arising from option-based trading strategies
aimed to express views about developments in swap rate volatility. These pitfalls motivate the
introduction of new interest rate variance contracts. Sections 4 and 5 develop the design, pric-
ing, hedging for these new contracts: Section 4 deals with contracts relying on percentage
volatility, i.e. the volatility of the logarithmic changes in the forward swap rate; Section 5 in-
troduces contracts on basis point or Gaussian volatility, related to absolute changes in the
forward swap rate. The contracts in Section 4 and 5 are priced in a model-free format, through
spanning arguments such as those in Bakshi and Madan (2000) and Carr and Madan (2001),
although, as noted already, the payos and contract designs we consider here are quite distinct
since they relate to xed income market volatility and pose new methodological issues. Our
contracts and indexes aim to match the market practice of quoting swaption implied volatil-
ity both in terms of percentage and basis point implied volatility. Section 6 suggests potential
trading strategies for the variance contracts introduced in Sections 4 and 5. Section 7 develops
indexes of swap rate volatility, which reect the fair value of the variance contracts of Sections
4 of 5. Section 8 explains in detail the main dierences between rate variance contracts and
indexes and those already in place in the equity case. Section 9 contains extensions where swap
markets can experience jumps, and show the remarkable property that even in these markets,
5 c by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
our basis point index is expressed in a model-free format. Section 10 uses sample market data
to illustrate a numerical implementation of one of the interest rate swap volatility indexes. An
appendix contains technical details.
2. Interest rate transactions risks
Consider a forward starting swap, i.e. a contract agreed at time t, starting at time 1, and
having : reset dates 1
0
. . 1
a1
and payment periods 1
1
1
0
. . 1
a
1
a1
, with 1 _ 1
0
,
summing up to a tenor period equal to 1
a
1
0
. Standard market practice sets 1 = 1
0
, as in
Figure 3, although this practice does not play any role for all the pricing and contract designs
put forward in this paper. Let 1
t
(1
1
. . 1
a
) be the forward swap rate prevailing at time t,
i.e. the xed rate such that the value of the forward starting swap is zero at t. In a swaption
contract agreed at time t, one counterparty has the right, but not the obligation, to enter a
swap contract at 1
0
< 1
1
, either as a xed rate payer or as a xed rate receiver, with the agreed
xed rate equal to some strike 1. If the swaption is exercized at 1, the value of the underlying
swap is, from the perspective of the xed rate payer:
SWAP
T
(1; 1
1
. . 1
a
) = PVBP
T
(1
1
. . 1
a
) [1
T
(1
1
. . 1
a
) 1] . (1)
where PVBP
T
(1
1
. . 1
a
) is the present value of receiving one dollar at each xed payment
date at the swaption expiry, also known as the swaps price value of the basis point(see, e.g.,
Brigo and Mercurio, 2006), i.e. the present value impact of one basis point move in the swap
rate at 1, 1
T
(1
1
. . 1
a
). Naturally, 1
t
(1
1
. . 1
a
) is both a forward and a spot swap rate
when t = 1. Appendix A contains a succinct summary of the evaluation framework underlying
swaption contracts, as well as the mathematical denitions of both the forward swap rate,
1
t
(1
1
. . 1
a
), and the price value of the basis point, PVBP
t
(1
1
. . 1
a
), (see Eqs. (A.1) and
(A.2)).
Swaptions can be priced through the Black (1976) formula once we assume the forward
swap rate has deterministic volatility. The standard market practice is to use this formula to
convert premiums to implied Black volatilities and vice-versa. There are two market conventions
for quoting implied volatilities: one in terms of percentage implied volatilities, similar to
equity options; and another in terms of basis point implied volatilities, which are percentage
volatilities multiplied by the current level of the forward swap rate. This paper develops variance
contracts and volatility indexes that match both strands of market practice.
An important feature of swaptions is that the underlying swap value is unknown until the
swaption expiration not only because forward swap rates uctuate but also because the PVBP
of the forward swaps xed leg also uctuates. This PVBP is the present value at time 1
of each dollar arising from paying the xed strike rate 1 rather than the realized swap rate
1
T
(1
1
. . 1
a
). It can be interpreted as a portfolio of zero coupon bonds maturing at the xed
payment dates of the swap, weighted by the lengths of the reset intervals and is determined by
the yield curve prevailing at time 1. Options on equities relate to a single source of risk, the
stock price. Swaps, instead, are tied to two sources of risk: (i) the forward swap rate, and (ii)
the swaps PVBP realized at time 1. This adds complexity to dening and pricing swap market
volatility.
6 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
3. Option-based volatility strategies
The standard instruments for trading interest rate swap volatility are swaption payers and
receivers. Consider the following strategies:
(a) A delta-hedged swaption position, i.e. a long position in a swaption, be it payer or receiver,
hedged through Blacks (1976) delta.
(b) A straddle, i.e. a long position in a payer and a receiver with the same strike, maturity,
and tenor.
Both strategies are analyzed in Appendix. In Appendix B, Eq. (B.9), we show the delta-
hedged swaption position leads, approximately, to the following daily P&L at the swaptions
expiry:
P&L
T
-
1
2
T

t=1

$
t

__
o
2
t
IV
2
0
_
PVBP
T

+
T

t=1
Track
t
Vol
t
(PVBP)

1
t
1
t
. (2)
where
$
t
is the Dollar Gamma, i.e. the swaptions Gamma times the square of the forward swap
rate, o
t
is the instantaneous volatility of the forward swap rate at day t, IV
0
is the swaption
implied percentage volatility at the time the strategy is rst implemented, Track
t
is the tracking
error of the hedging strategy, Vol
t
(PVBP) is the volatility of the PVBP rate of growth at t
and, nally,
1t
1t
denotes the series of shocks aecting the forward swap rate.
In Appendix B, Eq. (B.11), we show that a straddle strategy leads to a P&L similar to that in
Eq. (2), with the rst term being twice as that in Eq. (2), and with the straddle value replacing
the tracking error in the second term:
P&L
straddle
T
-
T

t=1

$
t

__
o
2
t
IV
2
0
_
PVBP
T

+
T

t=1
STRADDLE
t
Vol
t
(PVBP)

1
t
1
t
. (3)
where STRADDLE
t
is the value of the straddle as of time t. The advantage of this strategy
over the rst, is that it does not rely on hedging and, hence, it is not expensive in this respect.
However, it is striking to see how similar the P&L in Eqs. (2) and (3) are.
We emphasize that the P&L in Eqs. (2) and (3) are only approximations to the exact ex-
pressions given in Appendix B. For example, the exact version of the P&L in Eq. (3) has an
additional term, arising because trading straddles implies a delta that is only approximately
zero. It may well occur that in episodes of pronounced interest rate swap volatility, this delta
drifts signicantly away from zero. Naturally, not including this term in the approximations
of Eqs. (2) and (3) favorably biases the assessment of option-based based strategies to trade
interest rate swap volatility. This term would only add uncertainty to outcomes, on top of the
uncertainties singled out below.
Eqs. (2) and (3) show that the two option-based volatility strategies lead to quite unpre-
dictable prots that fail to isolate changes in interest rate swap volatility. There are two main
issues:
(i) For both strategies, the rst component of the P&L is proportional to the sum of the daily
P&L, given by the volatility view o
2
t
IV
2
0
, weighted with the Dollar Gamma,
$
t
, and the
7 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
PVBP at the swaption expiry. This is similar, but distinct (due to the PVBP at 1), from
what we already know from equity option trading (see Bossu, Strasser and Guichard, 2005,
and Chapter 10 in Mele, 2011). Indeed, one reason equity variance contracts are attractive
compared to options-based strategies is that they overcome the price dependency the
latter generates. This dependency shows up in our context as well: even when o
2
t
IV
2
0
0
for most of the time, it may happen too often that bad realizations of volatility, o
2
t
IV
2
0
<
0, occur precisely when the Dollar Gamma
$
t
is high. In other words, the rst terms in
Eq. (2) and Eq. (3) can lead to losses even if o
2
t
is higher than IV
2
0
for most of the time.
(ii) The second terms in Eqs. (2) and (3) show that option-based P&L depend on the shocks
driving the forward swap rate,
1t
1t
. In fact, even if future volatility is always higher
than the current implied volatility, o
2
t
IV
2
0
, the second terms in Eq. (2) and Eq. (3)
may actually dwarf the rst in the presence of adverse realizations of
1t
1t
. These second
terms in the P&L do not appear in option-based P&L for equity volatility. They are
specic to what swaptions are: instruments that have as underlying a swap contract,
whose underlying swap values PVBP is unknown until the maturity of the swaption, as
Eq. (1) shows. Uncertainty related to the PVBP, Vol
t
(PVBP) 0, cannot be hedged
away through the previous two option-based strategies.
These two issues lead to compelling theoretical reasons why options might not be appropriate
instruments to trade interest rate swap volatility. How reliable are option-based strategies for
that purpose in practice? Figure 4 plots the P&L of two directional volatility trading strategies
against the realized variance risk-premium, dened below, and calculated using daily data
from January 1998 to December 2009.
The two strategies are long positions in: (i) an at-the-money straddle; (ii) one of the interest
rate variance contracts introduced in the next section. Both positions relate to contracts expiring
in one and three months and tenors of ve years. The realized variance risk-premium is dened
as the dierence between the realized variance of the forward swap rate during the relevant
holding period (i.e. one or three months), and the squared implied at-the-money volatility
of the swaption (at one or three months) prevailing at the beginning of the holding period.
Appendix D provides all technical details regarding these calculations.
The top panel of Figure 4 depicts the two P&Ls relating to one-month trades (147 trades),
and the bottom panel displays the two P&Ls relating to three-month trades (49 trades). Do
these strategies deliver results consistent with directional views? In general, the P&Ls of both
the straddle-based strategy and the interest rate variance contract have the same sign as the
realized variance risk-premium. However, the straddle P&Ls are dispersed, in that they do
not always preserve the same sign of the realized variance risk-premium, a property displayed by
the interest rate variance contract. With straddles, the P&L has the same sign as the variance
risk-premium approximately 62% of the time for one-month trades, and for approximately 65%
of the time for three-month trades. The correlation between the straddle P&L and the variance
risk-premium is about 33% for the one-month trade, and about 28% for the three-month trade.
Further simulation studies performed by Jiang (2011) under a number of alternative holding
period assumptions suggest that if dynamically delta-hedged after inception, these trades lead
to higher correlationswith an average correlation over all the simulation experiments equal
to 63%. Instead, note the performance of the interest rate variance contract in our experiment,
which is a pure play in volatility, with its P&L lining up to a straight line and a correlation
with the variance risk-premium indistinguishable from 100%.
8 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
-0. 25 -0.2 -0. 15 -0.1 -0. 05 0 0.05 0.1 0.15 0.2
-0.1
-0. 08
-0. 06
-0. 04
-0. 02
0
0.02
0.04
0.06
Change in Forward Swap Rate Variance (1 month into 5 y ears)
P
&
L
s

o
f

V
o
l
a
t
i
l
i
t
y

C
o
n
t
r
a
c
t

a
n
d

S
t
r
a
d
d
l
e
One-month trade and contract
Volatility Contract
Straddle
-0.1 -0.05 0 0.05 0.1 0.15 0.2
-0.1
-0.05
0
0.05
0.1
0.15
Change in Forward Swap Rate Variance (3 months into 5 y ears)
P
&
L
s

o
f

V
o
l
a
t
i
l
i
t
y

C
o
n
t
r
a
c
t

a
n
d

S
t
r
a
d
d
l
e
Three-month trade and contract
Volatility Contract
Straddle
Figure 4. Empirical performance of directional volatility trades.
9 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
The reason straddles lead to more dispersed prots in the three-month trading period relates
to price dependency. For example, even if volatility is up for most of the time in a one month
period, it might be down during a given week, which inuences the P&L. However, straddles do
not weight these ups and downs in the same way. Rather, they might attribute a large weight
(given by the Dollar Gamma) to the particular week when volatility is down, as Eq. (3) makes
clear. The chances these unfortunate weightings happen increase with the duration of the trade.
While these facts are well-known in the case of equity volatility, the complexity of interest rate
transactions makes these facts even more severe when it comes to trading interest rate swap
volatility through options.
4. Interest rate swap variance contracts
We assume that the forward swap rate 1
t
(1
1
. . 1
a
) is a diusion process with stochastic
volatility, and consider extensions to a jump-diusion case in Section 9. It is well-known that
the forward swap rate is a martingale under the so-called swap probability, such that in a
diusive environment,
d ln 1
t
(1
1
. . 1
a
) =
1
2
|o
t
(1
1
. . 1
a
)|
2
dt +o
t
(1
1
. . 1
a
) d\
+
t
. t [t. 1] . (4)
where \
+
t
is a multidimensional Wiener process under the swap probability, and o
t
(1
1
. . 1
a
)
is adapted to \
+
t
. Accordingly, dene \
a
(t. 1), as the realized variance of the forward swap
rate logarithmic changes in the time interval [t. 1], for tenor length 1
a
1:
\
a
(t. 1) =
_
T
t
|o
t
(1
1
. . 1
a
)|
2
dt. (5)
Appendix A summarizes additional notation and basic assumptions and facts about forward
starting swaps, swap rates, the swap probability, and the pricing of swaptions.
This section designs contracts to trade the forward swap rate volatility, \
a
(t. 1), that over-
come the issues inherent in option trading as observed in Section 3. The contracts are priced
in a model-free fashion, and lead to model-free indexes of expected volatility over a reference
period [t. 1]introduced in Section 7. This section considers contracts referencing percentage,
or logarithmic variance, and Section 5 develops their basis point counterparts.
4.1. Contracts
We consider three contracts: a forward agreement, which requires an initial payment, and two
variance swaps, which are settled at the expiration. These contracts are all equally useful for
the purpose of implementing views about swap market volatility developments, as explained in
Section 6.
We begin with the forward agreement. Our rst remark is that uncertainty about swap
markets relates to the volatility of the forward swap rate, rescaled by the PVBP, as Eq. (1)
reveals, with the complication that the PVBP is unknown until 1. Consider the denition of
the realized variance of the forward swap rate, \
a
(t. 1) in Eq. (5), and the following forward
agreement:
10 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
Denition I (Interest Rate Variance Forward Agreement). At time t, coun-
terparty Apromises to pay counterparty Bthe product of forward swap rate variance
realized over the interval [t. 1] times the swaps PVBP prevailing at time 1, i.e. the
value \
a
(t. 1) PVBP
T
(1
1
. . 1
a
). The price counterparty B shall pay counter-
party A for this agreement at time t is called the Interest Rate Variance (IRV)
Forward rate, and is denoted as F
var,a
(t. 1).
The PVBP is the price impact of a one basis point move in the swap rate over the tenor at
time 1, which is unknown before time 1, as further discussed after Denition III. Rescaling by
the forward PVBP is mathematically unavoidable when the objective is to price volatility in
a model-free fashion. Our goal is to simultaneously price interest rate swap volatility, swaps,
swaptions and pure discount bonds in such a way that the price of volatility conveys all the
information carried by all the traded swaptions and bonds. Eq. (1) suggests that the uncertainty
related to the interest rate swaps underlying the swaptions is tied to both the realized variance,
\
a
(t. 1), and the forward PVBP. To insulate and price volatility, the payo of the variance
contract needs to be rescaled by PVBP
T
(1
1
. . 1
a
), just as the value of the swap does in
Eq. (1). Note that the term PVBP
T
(1
1
. . 1
a
) also appears in the P&L of the swaption-
based volatility strategies, as shown by Eqs. (2) and (3); the important dierence is that the
variance contract underlying Denition I isolates volatility from Dollar Gamma.
1
Finally, from
a practical perspective, such a rescaling does not aect the ability of the contract to convey
views about future developments of \
a
(t. 1): as noted, the correlation between the variance
risk-premium and the P&L of IR-variance contracts is nearly 100% in the empirical experiment
in Figure 4.
In fact, the P&L in Figure 4 refer to a variance contract cast in a swap format, whereby
counterparty A agrees to pay counterparty B the following dierence at time 1:
Var-Swap
a
(t. 1) = \
a
(t. 1) PVBP
T
(1
1
. . 1
a
) P
var,a
(t. 1) . (6)
where P
var,a
(t. 1) is a xed variance swap rate, determined at time t, chosen in a way to lead
to the following zero value condition at time t:
Denition II (IRV Swap Rate). The IRV swap rate is the xed variance swap
rate P
var,a
(t. 1), which makes the current value of Var-Swap
a
(t. 1) in Eq. (6) equal
to zero.
Note that the variance contract arising from Denition II is a forward, not really a swap.
However, we utilize a terminology similar to that already in place for equity variance contracts,
and simply refer to the previous contract as a swap.
Finally, we consider a second denition of the IRV swap rate, which will lead to a denition
of the index of interest rate swap volatility in Section 7. Consider the following payo:
Var-Swap
+
a
(t. 1) =
_
\
a
(t. 1) P
+
var,a
(t. 1)

PVBP
T
(1
1
. . 1
a
) . (7)
where P
+
var,a
(t. 1), a xed variance swap rate determined at t, is set so as to lead to a zero value
condition at time t:
1
If the risk-neutral expectation of the short-term rate is insensitive to changes in the short-term swap rate volatility, the PBVT
at time T is increasing in the short-term swap rate volatility (see Mele, 2003). In this case, the IRV forward agreement is a device
to lock-in both volatility of the forward swap rate from t to T, and short-term swap rate volatility at T.
11 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
Denition III (Standardized IRV Swap Rate). The Standardized IRV swap
rate is the xed variance swap rate P
+
var,a
(t. 1) that makes the current value of
Var-Swap
+
a
(t. 1) in Eq. (7) equal to zero.
The payo in Eq. (7) carries an intuitive meaning. It is the product of two terms: (i) the dif-
ference between the realized variance and a fair strike and (ii) the PVBP at time 1. This payo
is similar to that of standard equity variance contracts: the holder of the contract would receive
PVBP
T
(1
1
. . 1
a
) dollars for every point by which the realized variance exceeds the variance
strike price. The added complication of the IRV contract design is that PVBP
T
(1
1
. . 1
a
) is
unknown at time t.
4.2. Pricing
In the absence of arbitrage and market frictions, the price of the IRV forward agreement to be
paid at time t, F
var,a
(t. 1) in Denition I, is expressed in a model-free format as a combination of
the prices of at-the-money and out-of-the-money swaptions. One approximation to F
var,a
(t. 1)
based on a nite number of swaptions, is, keeping the same notation:
F
var,a
(t. 1) = 2
_

i:1
i
<1t
SWPN
R
t
(1
i
. 1; 1
a
)
1
2
i
1
i
+

i:1
i
1t
SWPN
P
t
(1
i
. 1; 1
a
)
1
2
i
1
i
_
(8)
where SWPN
R
t
(1
i
. 1; 1
a
) (resp., SWPN
P
t
(1
i
. 1; 1
a
)) is the price of a swaption receiver (resp.,
payer), struck at 1
i
, expiring at 1 and with tenor extending up to time 1
a
, and 1
i
=
1
2
(1
i+1
1
i1
) for 1 _ i < `, 1
0
= (1
1
1
0
), 1
A
= (1
A
1
A1
), where 1
0
and
1
A
are the lowest and the highest available strikes traded in the market, and ` + 1 is the
total number of traded swaptions expiring at time 1 and with tenor extending up to time
1
a
. Appendix C provides the theoretical framework underlying Eq. (8). Appendix E provides
an expression for the local volatility surface, which we can use to interpolate the missing
swaptions from those that are traded, so as to ll in Eq. (8).
In Appendix C, we show that the swap rate in Denition II is given by:
P
var,a
(t. 1) =
F
var,a
(t. 1)
1
t
(1)
(9)
where 1
t
(1) is the price of a zero coupon bond maturing at 1 and time-to-maturity 1 t.
Finally, the standardized swap rate in Denition III is:
P
+
var,a
(t. 1) =
F
var,a
(t. 1)
PVBP
t
(1
1
. . 1
a
)
(10)
4.3. Hedging
The interest rate variance contracts can be hedged using swaptions underlying the evaluation
framework of Section 4.2. An IRV market maker will be concerned with the replication of the
payos needed to hedge the contracts. In the exposition below, we take the perspective of a
provider of insurance against volatility who sells any of the IRV contracts underlying Denitions
I through III of Section 4.1.
12 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
The IRV forward contract in Denition I can be hedged through the positions in rows (i)
and (ii) of Tables I and II, which we now discuss. Next, consider the contracts relating to
the IRV swap rate of Denition II, P
var,a
(t. 1). We aim to replicate the payo in Eq. (6). As
shown in Appendix C, we hedge against this payo through two, identical, zero cost portfolios,
constructed with the following positions:
(i) A dynamic position in a self-nanced portfolio of zero coupon bonds aiming to replicate
the cumulative increments of the forward swap rate over [t. 1], which turns out to equal
_
T
t
o1s(T
1
, ,Tn)
1s(T
1
, ,Tn)
=
1
2
\
a
(t. 1) + ln
1
T
(T
1
, ,Tn)
1t(T
1
, ,Tn)
, rescaled by the PVBP at time 1.
(ii) Static positions in swaps starting at time 1 and OTMswaptions expiring at time 1, aiming
to replicate the payo of the interest rate counterpart to the log-contract (Neuberger,
1994) for equities, rescaled by the PVBP at 1. The positions are as follows:
(ii.1) Short 1,1
t
(1
1
. . 1
a
) units of a forward starting (at 1) xed interest payer swap
struck at the current forward swap rate, 1
t
(1
1
. . 1
a
).
(ii.2) Long out-of-the-money swaptions, each of them carrying a weight equal to
1
i
1
2
i
,
where 1
i
and 1
i
are as in Eq. (8).
(iii) A static borrowing position aimed to nance the out-of-the-money swaption positions in
(ii.2).
The details for the self-nanced portfolio in (i) are given in Appendix C. Table I summarizes
the costs of each of these trades at t, as well as the payos at 1, which sum up to be the same
as the payo in Eq. (6), as required to hedge the contract.
Table I
Replication of the contract relying on the IRV swap rate of Denition II. ZCB and OTM
stand for zero coupon bonds and out-of-the-money, respectively, and 1
t
= 1
t
(T
1
, , T
a
),
IVLI
T
= IVLI
T
(T
1
, , T
a
).
Portfolio Value at t Value at T
(i) long self-nanced portfolio of ZCB 0
_
\
a
(t, T) 2 ln
1
T
1t
_
IVLI
T
(ii) short swaps and long OTM swaptions F
var,a
(t, T) 2 ln
1
T
1t
IVLI
T
(iii) borrow F
var,a
(t, T) F
var,a
(t, T)
Fvar;n(t,T)
1t(T)
= P
var,a
(t, T)
Net cash ows 0 \
a
(t, T) IVLI
T
P
var,a
(t, T)
Finally, to hedge against the Standardized contract of Denition III, we need to replicate
the payo in Eq. (7). We create a portfolio, which is the same as that in Table I, except
now that row (iii) refers to a borrowing position in a basket of zero coupon bonds with
value equal to PVBP
t
(1
1
. . 1
a
) and notional of P
+
var,a
(t. 1), as in Table II. By Eq. (10),
the borrowing position needed to nance the out-of-the-money swaptions in row (ii), is just
F
var,a
(t. 1) = P
+
var,a
(t. 1) PVBP
t
(1
1
. . 1
a
). Come time 1, it will be closed for a value of
P
+
var,a
(t. 1) PVBP
T
(1
1
. . 1
a
). The net cash ows are precisely those of the contract with
the Standardized IRV swap rate.
13 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
Table II
Replication of the contract relying on the Standardized IRV swap rate of Denition III.
ZCB and OTM stand for zero coupon bonds and out-of-the-money, respectively, and
1
t
= 1
t
(T
1
, , T
a
), IVLI
t
= IVLI
t
(T
1
, , T
a
).
Portfolio Value at t Value at T
(i) long self-nanced portfolio of ZCB 0
_
\
a
(t, T) 2 ln
1
T
1t
_
IVLI
T
(ii) short swaps and long OTM swaptions F
var,a
(t, T) 2 ln
1
T
1t
IVLI
T
(iii) borrow basket of ZCB for P

var;n
(t, T) PVBPt F
var,a
(t, T) P
+
var,a
(t, T) IVLI
T
Net cash ows 0
_
\
a
(t, T) P
+
var,a
(t, T)

IVLI
T
The replication arguments in this section are alternative means to determine the no-arbitrage
value of the IRV forward agreement of Denition I, and the IRV swap rates in Denitions II and
III, P
var,a
(t. 1) and P
+
var,a
(t. 1). Suppose, for example, that the market value of the Standardized
IRV swap rate, P
+
var,a
(t. 1)
$
say, is higher than the no-arbitrage value P
+
var,a
(t. 1) in Eq. (10).
Then, one could short the standardized contract underlying Denition III and, at the same
time, synthesize it through the portfolio in Table II. This positioning costs zero at time t and
yields a sure positive prot at time 1 equal to
_
P
+
var,a
(t. 1)
$
P
+
var,a
(t. 1)

PVBP
T
. To rule
out arbitrage, then, we need to have P
+
var,a
(t. 1)
$
= P
+
var,a
(t. 1).
5. Basis point volatility
5.1. Contracts and evaluation
We can price and hedge volatility based on arithmetic, or basis point (BP henceforth), changes
of the forward swap rate in Eq. (4),
d1
t
(1
1
. . 1
a
) = 1
t
(1
1
. . 1
a
) o
t
(1
1
. . 1
a
) d\
+
t
. t [t. 1] . (11)
Accordingly, denote the BP realized variance with \
BP
a
(t. 1),
\
BP
a
(t. 1) =
_
T
t
1
2
t
(1
1
. . 1
a
) |o
t
(1
1
. . 1
a
)|
2
dt (12)
We can then restate the interest rate variance contracts in Section 4.1 in terms of BP variance
and, for reasons explained below, we refer to the ensuing contracts as Gaussian contracts,
dened as follows:
Denition IV (Gaussian Contracts). Consider the following contracts and swap
rates replacing those in Denitions I, II, and III:
(a) (BP-IRV Forward Agreement). At time t, counterparty A promises to pay coun-
terparty B the product of the forward swap rate BP variance realized over the interval
[t. 1] times the swaps PVBP that will prevail at time 1, i.e. the value \
BP
a
(t. 1)
PVBP
T
(1
1
. . 1
a
). The price counterparty B shall pay counterparty A for this agree-
ment at time t is called BP-IRV Forward rate, and is denoted as F
BP
var,a
(t. 1).
14 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
(b) (BP-IRVSwap Rate). The BP-IRVswap rate is the xed variance swap rate P
BP
var,a
(t. 1),
which makes the current value of \
BP
a
(t. 1) PVBP
T
(1
1
. . 1
a
) P
BP
var,a
(t. 1) equal to
zero.
(c) (Standardized BP-IRV Swap Rate). The Standardized BP-IRV swap rate is the xed
variance swap rate P
+BP
var,a
(t. 1), which makes the current value of
_
\
BP
a
(t. 1) P
+BP
var,a
(t. 1)

PVBP
T
(1
1
. . 1
a
) equal to zero.
The reason we refer to these contracts as Gaussian relates to the fact that a benchmark
case is one where the realized BP variance of the forward swap rate in Eq. (11) is constant and
equal to o
.
,
d1
t
(1
1
. . 1
a
) = o
.
d\
+
t
. t [t. 1] . (13)
such that the forward swap rate has a Gaussian distribution. We shall return to this abstract
assumption of a constant BP variance below to develop intuition about the pricing results
relating to the general stochastic BP variance case as dened in Eq. (12).
In Appendix G, we show that in the general case of Eqs. (11) and (12), the price of the
BP-IRV forward contract is approximated by:
F
BP
var,a
(t. 1) = 2
_

i:1
i
<1t
SWPN
R
t
(1
i
. 1; 1
a
) 1
i
+

i:1
i
1t
SWPN
P
t
(1
i
. 1; 1
a
) 1
i
_
(14)
where the strike dierences, 1
i
, are as in Eq. (8). Given the BP-IRV forward, the BP-IRV
swap rates in Denition IV-(a) and Denition IV-(b) are:
P
BP
var,a
(t. 1) =
F
BP
var,a
(t. 1)
1
t
(1)
and P
+BP
var,a
(t. 1) =
F
BP
var,a
(t. 1)
PVBP
t
(1
1
. . 1
a
)
(15)
The intuition behind the forward price in Eq. (14) is that the instantaneous BP-variance is
simply the instantaneous variance of the logarithmic changes of the forward swap rate rescaled
by the forward swap rate, as in Eq. (11), and squared. This property translates into an analogous
property for the price of the forward variance agreement: comparing Eq. (8) with Eq. (14)
reveals that for the BP agreement, each swaption price i carries the same weight as that for
the agreement in Denition I,
1
i
1
2
i
, rescaled by the squared strike 1
2
i
, i.e.
1
i
1
2
i
1
2
i
.
In their derivation of the fair value of equity variance swaps, Demeter, Derman, Kamal and
Zou (1999) develop an intuitive approach relying on the Black and Scholes (1973) market, and
explain that a portfolio of options has a vega that is insensitive to changes in the stock price
only when the options are weighted inversely proportional to the square of the strike, which
also relates to the fair value of our forward variance agreement in Eq. (8). We develop a similar
approach to gain intuition about the reasons underlying the uniform swaptions weightings in
Eq. (14).
Assume a Gaussian market, i.e. one where the forward swap rate is as in Eq. (13). This
assumption is the obvious counterpart to that of a constant percentage volatility underlying
the standard Black-Scholes market for equity options. Denote with O
t
(1. 1. 1. o
.
; 1
a
) the
price of a swaption (be it a receiver or a payer) at time t when the forward swap rate is 1.
We create a portfolio with a continuum of swaptions having the same maturity and tenor, and
15 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
denote with . (1) the portfolio weigthings, assumed to be independent of 1, and such that the
value of the portfolio is:
:
t
(1
t
(1
1
. . 1
a
) . o
.
) =
_
. (1) O
t
(1
t
(1
1
. . 1
a
) . 1. 1. o
.
; 1
a
) d1.
We require that the vega of the portfolio, dened as i
t
(1. o) =
0t(1,o)
0o
, be insensitive to
changes in the forward swap rate,
Ji
t
(1. o)
J1
= 0. (16)
In Appendix G, we show that the vega of the portfolio does not respond to 1 if and only if the
weightings, . (1), are independent of 1,
Eq. (16) holds true ==. (1) = const. (17)
That is, a portfolio of swaptions aiming to track BP variance, thereby being immuned to changes
in the underlying forward swap rate, is one where the swaptions are equally weighted.
5.2. Hedging
Replicating BP-denominated contracts requires positioning in a way quite dierent from repli-
cating percentage volatility as explained in Section 4.3. As noted in the previous section, the
hedging arguments summarized in Tables I and II rely on the replication of log-contracts, i.e.
contracts delivering ln(
1
T
1t
), rescaled by the PVBP. Instead, to hedge BP contracts, we need
to rely on quadratic contracts, i.e. those delivering 1
2
T
1
2
t
, rescaled by the PVBP, as we now
explain.
Consider, rst, the payo of the BP-IRV forward contract in Denition IV-(a), which can be
replicated through the portfolios in rows (i) and (ii) of Tables III and IV, as further discussed
below. To replicate the contract for the BP-IRV swap rate of Denition IV-(b), we create
portfolios comprising the following positions:
(i) A dynamic, short position in two identical, self-nanced portfolios of zero coupon bonds
that replicate the increments of the forward swap rate over [t. 1], weighted by the forward
swap rate, which turns out to equal
_
T
t
1
c
(1
1
. . 1
a
) d1
c
(1
1
. . 1
a
) =
1
2
_
\
BP
a
(t. 1) (1
2
T
(1
1
. . 1
a
) 1
2
t
(1
1
. . 1
a
))

, rescaled by the PVBP at time 1.

(ii) Static positions in swaps starting at time 1 and OTM swaptions expiring at time 1,
aiming to replicate the payo of a quadratic contract, rescaled by the PVBP at 1. The
positions are as follows:
(ii.1) Long 21
t
(1
1
. . 1
a
) units of a forward starting (at 1) xed interest payer swap
struck at the current forward swap rate, 1
t
(1
1
. . 1
a
).
(ii.2) Long out-of-the-money swaptions, each of them carrying a weight equal to 21
i
,
where 1
i
and 1
i
are as in Eq. (8).
(iii) A static borrowing position aimed to nance the out-of-the-money swaption positions in
(ii.2).
Costs and payos of these portfolios are summarized in Table III. The result is a perfect
replication of the payo relating to the contract of Denition IV-(b).
16 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
Table III
Replication of the payo relating to the BP-IRV contract of Denition IV-(b). ZCB
and OTM stand for zero coupon bonds and out-of-the-money, respectively, and 1
t
=
1
t
(T
1
, , T
a
), IVLI
T
= IVLI
T
(T
1
, , T
a
).
Portfolio Value at t Value at T
(i) short self-nanced portfolio of ZCB 0
_
\
BP
n
(t, T)
_
1
2
T
1
2
t
__
PVBP
T
(ii) long swaps and long OTM swaptions F
BP
var,a
(t, T)
_
1
2
T
1
2
t
_
IVLI
T
(iii) borrow F
BP
var,a
(t, T) F
BP
var,a
(t, T)
F
BP
var;n
(t,T)
1t(T)
= P
BP
var,a
(t, T)
Net cash ows 0 \
BP
n
(t, T) PVBP
T
P
BP
var;n
(t, T)
Finally, Table IV is obtained by modifying the arguments leading to Table III, in the same
way as we did to obtain Table II in Section 4.3.
Table IV
Replication of the payo relating to the Standardized BP-IRV contract of Denition IV-
(c). ZCB and OTM stand for zero coupon bonds and out-of-the-money, respectively, and
1
t
= 1
t
(T
1
, , T
a
), IVLI
t
= IVLI
t
(T
1
, , T
a
).
Portfolio Value at t Value at T
(i) short self-nanced portfolio of ZCB 0
_
\
BP
n
(t, T)
_
1
2
T
1
2
t
__
PVBP
T
(ii) long swaps and long OTM swaptions F
BP
var,a
(t, T)
_
1
2
T
1
2
t
_
IVLI
T
(iii) borrow basket of ZCB for P
BP
var;n
(t, T) PVBPt F
BP
var,a
(t, T) P
+BP
var,a
(t, T) IVLI
T
Net cash ows 0
_
\
BP
n
(t, T) P
BP
var;n
(t, T)
_
PVBP
T
5.3. Links to constant maturity swaps
We show that a constant maturity swap (CMS) can be represented as a basket of the BP-IRV
forwards of Denition IV-(a), suggesting interest rate variance swaps can be used to hedge
against CMS. The technical reason IRV forwards link to CMS is due to a classical convexity
adjustment of the CMS, arising because the price of a CMS relates to the wrong expectation
of future payos, one under which the forward swap rate is not a martingale.
In a standard constant maturity swap, a party pays a counterparty the spot swap rate with a
xed tenor over a sequence of dates, with legs set in advance, i.e.
1
T
0
+)i
(1
1
+,i. . 1
a
+,i) at times 1
0
+ (, + 1) i, , = 0. . ` 1, where i is a xed
constant, for example
1
2
, and ` is the number of payment dates. In Appendix G, we show that
the price of a CMS is approximately equal to:
CMS
.
(t) =
.1

)=0
1
t
(t
)
+/) 1
t
(1
1
+,i. . 1
a
+,i)
+
.1

)=0
G
t
(1
t
(1
1
+,i. . 1
a
+,i)) F
BP
var,a
(t. 1
0
+,i) . (18)
where F
BP
var,a
(t. 1) is the price of the BP-IRV forward contract in Denition IV-(a), as given by
Eq. (14), and G() is a function given in the Appendix (see Eq. (G.8)).
17 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
That the price of a CMS relates to the entire swaption skew is known at least since Hagan
(2003) and Mercurio and Pallavicini (2006), as further discussed in Appendix G. As is also clear
from Eq. (18), CMS can be hedged through a basket of IRV forwards in BP expiring at the
points preceding the CMS payments.
6. Marking to market and trading strategies
This section provides expressions for marking to market updates of the IRV contracts of Sections
4 and 5 (see Section 6.1). These expressions also help assess the value of trading strategies based
on these contracts, as explained in Section 6.2.
6.1. Marks to market of IRV swaps
In Appendix C, we show that the value at any time t (t. 1), of the variance contract struck
at time t at the IRV swap rate P
var,a
(t. 1) in Eq. (9) is:
M-Var
a
(t. t. 1) = \
a
(t. t) PVBP
t
(1
1
. . 1
a
) 1
t
(1) [P
var,a
(t. 1) P
var,a
(t. 1)] . (19)
and the value of the contract struck at the Standardized IRV swap rate P
+
var,a
(t. 1) in Eq. (10)
is:
M-Var
+
a
(t. t. 1) = PVBP
t
(1
1
. . 1
a
)
_
\
a
(t. t)
_
P
+
var,a
(t. 1) P
+
var,a
(t. 1)
_
. (20)
It is immediate to see that the marking to market updates for Basis Point contracts are
the same as those in Eqs. (19) and (20)diering only by a mere change in notation. In the
next section, we utilize these marking to market expressions to calculate the value of trading
strategies.
6.2. Trading strategies
We consider examples of trading strategies relying on the IRV contracts of Sections 4 and 5.
We analyze strategies relying on views about developments of: (i) spot IRV swap rates relative
to current (in Section 6.2.1), and (ii) the value of spot IRV swap rates relative to forward IRV
swap ratesi.e. those implied by the current term-structure of spot IRV swap rates (in Section
6.2.2). The strategies we consider apply to both percentage and BP contracts, although to
simplify the presentation, we only illustrate the case applying to percentage contracts.
6.2.1. Spot trading
[Link]. Trading through IRV swap rates
First, we suggest how to express views on IRV contracts that have dierent maturities and
tenors. Let : =
1
12
, and consider, for example, Figure 5, which illustrates the timing of two
contracts: (i) a contract expiring in 3 months, relating to the volatility of 3: into
^
1
5
3:
forward swap rates; and (ii) a contract expiring in 9 months, relating to the volatility of 9:
into 1
5
9: forward swap rates. Time units are expressed in years, and we set t = 0. We
assume the two contracts have the same tenor,
^
1
5
3: = 1
5
9:.
18 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
t=0

T=3m T=9m

T
2

T
5

T
1
T
2
T
5
T
Figure 5. Rolling tenor trade.
Let P
var
(t. 1) and P
var
(t. 1) be the IRV swap rates of these two contracts, and suppose
we have the view that at some point t 3:,
P
var
(t. 3:) P
var
(0. 3:) and P
var
(t. 9:) < P
var
(0. 9:) . (21)
A trading strategy consistent with these views might be as follows: (i) go long the 3:contract,
(ii) short c units of the 9: contract, (iii) short the IRV forward contract in Denition I for the
3: maturity, with contract length equal to t, (iv) go long c units of the IRV forward contract
for the 9: maturity, with contract length equal to t. The value of this strategy at time t is,
by marking-to-market according to Eq. (19):
M-Var

(0. t. 3:) c M-Var

(0. t. 9:)
\

(0. t) PVBP

1
1
. .

1

) +c \

(0. t) PVBP

(1
1
. . 1

)
= 1

(3:) [P
var
(t. 3:) P
var
(0. 3:)] +c 1

(9:) [P
var
(0. 9:) P
var
(t. 9:)]
0.
Its cost at t = 0 is F
var
(0. 3:) +c F
var
(0. 9:). We choose,
c =
F
var
(0. 3:)
F
var
(0. 9:)
. (22)
to make the portfolio worthless at t = 0.
The technical reason to enter into the IRV forward contract is that the time t value of the
variance contracts (i) and (ii) entails, by Eq. (19), cumulative realized variance exposures equal
to \

(0. t) PVBP

1
1
. .

1

) and \

(0. t) PVBP

(1
1
. . 1

), which cannot oset,

as they refer to two distinct rates: the 3: into

1
5
3: forward swap rate and the 9: into
1
5
9: forward swap rate. The device to short c units of the 9: contract serves the purpose
of osetting these two exposures at time t, and the expression for c in Eq. (22) allows us to
create a zero-cost portfolio at t = 0.
Next, we develop an example of a trading strategy relying on contracts with dierent matu-
rities but the same forward swap rates. Assume that at time t 0, the term structure of the
IRV swap rate P
var
(0. 1) is increasing in 1 for some xed tenor, but that a view is held that
this term structure is about to atten soon. For example, we might expect that at some point
in time t,
P
var
(t. 3:) P
var
(0. 3:) and P
var
(t. 9:) < P
var
(0. 9:) . (23)
where t 3:.
To express this view, one could go long a 3 month IRV swap struck at P
var
(0. 3:) and
shorts a 9 month IRV swap struck at P
var
(0. 9:). Note that this trading strategy relies on
19 c by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
IRV swaps for forward swap rates that are 9 months into 1

9: years, at time t = 0, as
Figure 6 illustrates.
t=0 T1 Tn

T2
T
0
=9m T=3m

Figure 6. Fixed tenor trade.

The payos to these two IRV swaps are as follows:
(i) The payo of the 3 month IRV swap at 1 = 3: is \

(0. 3:) PVBP

3
(1
1
. . 1

),
where the PVBP
3
(1
1
. . 1

) relates to swaps payments starting at time 1

1
.
(ii) The payo of the 9 month IRV swap at 1 = 9: is \

(0. 9:) PVBP

9
(1
1
. . 1

), and
PVBP
9
(1
1
. . 1

) relates to swaps payments starting at time 1

1
.
The price of the 9 month contract relies on the price of out-of-the-money swaptions that are
9 months into 1

9: years, as of time t = 0. The price of the 3 month contract is instead

calculated through the price of out-of-the-money swaptions with expiry 1 = 3: and underlying
swap starting at 1
0
= 9:. Typically, swaption markets have 1 = 1
0
, although theoretically, the
pricing framework in this section is still valid even when 1 < 1
0
as it is for the 3 month contract
in Figure 6. Note that markets may not exist or, at best, be illiquid, for the swaptions needed
to price volatility products traded for an expiry trade occurring prior to the beginning of the
tenor period of the underlying forward swap rate, 1 < 1
0
. In this case, the pricing of these
products could not be model-free. The simplest solution to price these illiquid assets might rely
on the Blacks (1976) formula, where one uses as an input the implied volatilities for maturity
1
0
t and replaces, everywhere in the formula, time to maturity 1
0
t with time to maturity
1 t. Note that we could simply not use swaptions with time to maturity 1 t to price these
products, if the swaps and, hence, the forward swap rates underlying these swaptions start exist
at time 1, not at time 1
0
as required in Figure 6.
The strategy of Figure 6 eliminates the risk related to the uctuations of realized interest
rate swap volatility, as we now explain. Come time t, its value is:
M-Var

(0. t. 3:) M-Var

(0. t. 9:)
= \

(0. t) PVBP

(1
1
. . 1

) +1

(3:) [P
var
(t. 3:) P
var
(0. 3:)]
(\

(0. t) PVBP

(1
1
. . 1

) +1

(9:) [P
var
(0. 9:) P
var
(t. 9:)])
0.
where we have made use of Eq. (19), and the last inequality follows by the view summarized by
Eqs.(23). Note that the strategy eliminates the cumulative realized variance exposure \

(0. t)
PVBP

(1
1
. . 1

), as shown by the previous calculations, precisely because we are trading

the two IRV swaps in Figure 6, which refer to the volatility of the same forward swap rate at
t = 0: the forward swap rate of 9 months into 1

9: years. As a result, a xed tenor trade

does not need additional positions in the IRV forward contract, as in the case of the rolling
tenor trades in Figure 5.
20 c by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
[Link]. Trading through Standardized IRV swap rates
While the trading strategies of the previous section rely on the IRV swap rate of Denition II,
trading the Standardized IRV swap rate of Denition III leads to comparable outcomes. For
example, suppose that, in analogy with Eq. (21), the view is held that, at some future point
t _ 3:,
P
+
var,^ a
(t. 3:) P
+
var,^ a
(0. 3:) and P
+
var,a
(t. 9:) < P
+
var,a
(0. 9:) .
where the dynamics of the contracts are, now, as in Figure 5. Consider the following strategy:
(i) go long the 3: contract, (ii) short c units of the 9: contract, (iii) short the IRV forward
contract of Denition I for the 3: maturity, with contract length equal to t, (iv) go long c
units of the IRV forward contract for the 9: maturity, with contract length equal to t. Using
the marking to market updates in Eq. (20), we nd that the value of this strategy at time t is:
M-Var
+
^ a
(0. t. 3:) c M-Var
+
a
(0. t. 9:)
\
^ a
(0. t) PVBP
t
(
^
1
1
. .
^
1
a
) +c \
a
(0. t) PVBP
t
(1
1
. . 1
a
)
= PVBP
t
(
^
1
1
. .
^
1
a
)
_
P
+
var,^ a
(t. 3:) P
+
var,^ a
(0. 3:)

+ PVBP
t
(1
1
. . 1
a
)
_
P
+
var,a
(0. 9:) P
+
var,a
(t. 9:)

0.
where c is as in Eq. (22), which ensures the portfolio is worthless at t = 0.
Likewise, suppose that, similarly as in Eq. (23), the view is that, at some future point t _ 3:,
P
+
var,a
(t. 3:) P
+
var,a
(0. 3:) and P
+
var,a
(t. 9:) < P
+
var,a
(0. 9:) .
where the dynamics of the contracts are exactly as in Figure 6. We go long the 3: and short
the 9: Standardized contracts. By Eq. (20), the value of this strategy at time t is:
M-Var
+
a
(0. t. 3:) M-Var
+
a
(0. t. 9:)
= PVBP
t
(1
1
. . 1
a
)
__
P
+
var,a
(t. 3:) P
+
var,a
(0. 3:)
_
+
_
P
+
var,a
(0. 9:) P
+
var,a
(t. 9:)
_
0.
6.2.2 Forward trading
Finally, we consider examples of trading strategies to implement views about the implicit pricing
that we can read through the current IRV swap rates. Suppose, for example, that at time t = 0,
we hold the view that in one year time, the IRV swap rate for a contract expiring in a further
year will be greater than the implied forward IRV swap rate, i.e. the rate implied by the
current term structure of the IRV contracts, viz
P
var,a
(1. 2) P
var,a
(0. 2) P
var,a
(0. 1) . (24)
The three IRV swap rates refer to contracts with tenors starting in two year time from t = 0,
and exhausting at the same time 1
a
, similarly as for the contracts underlying the xed tenor
trades in Figure 6.
We synthesize the cheap, implied IRV swap rate, by going long the following portfolio at
t = 0:
(i) long a two year IRV swap, struck at P
var,a
(0. 2)
(ii) short a one year IRV swap, struck at P
var,a
(0. 1)
[P.1]
21 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
This portfolio is costless at time zero, and if Eq. (24) holds true in one year, then in one
years time we close (i) and access the payo in (ii), securing a total payo equal to:
: = M-Var
a
(0. 1. 2) + [P
var,a
(0. 1) \
a
(0. 1) PVBP
1
(1
1
. . 1
a
)] .
The rst term is the market value of the long position in (i) in one years time, with M-
Var
a
(0. 1. 2) as in Eq. (19). The second term is the payo arising from (ii). Using Eq. (19), we
obtain,
: = [P
var,a
(1. 2) P
var,a
(0. 2)] 1
1
(2) +P
var,a
(0. 1)
_ [P
var,a
(1. 2) P
var,a
(0. 2) +P
var,a
(0. 1)] 1
1
(2)
0.
where the rst inequality follows by 1
1
(2) < 1, and the second holds by Eq. (24).
Likewise, we can implement trading strategies consistent with views about implied Standard-
ized IRV swap rates. Suppose we anticipate that:
P
+
var,a
(1. 2) P
+
var,a
(0. 2) P
+
var,a
(0. 1) .
where the dynamics of the contracts are again as in Figure 6. Consider the same portfolio as in
[P.1] with Standardized IRV contracts replacing IRV contracts. In one years time, the value of
this portfolio is:
M-Var
+
a
(0. 1. 2) +
_
P
+
var,a
(0. 1) \
a
(0. 1)

PVBP
1
(1
1
. . 1
a
)
=
_
P
+
var,a
(1. 2) +P
+
var,a
(0. 1) P
+
var,a
(0. 2)

PVBP
1
(1
1
. . 1
a
)
0.
where we have used the marking to market update in Eq. (20).
7. Swap volatility indexes
7.1. Percentage
The price of the IRV forward agreement of Denition I can be normalized by the length of the
contract, 1 t, and quoted in terms of the current swaps PVBP, so as to lead to the following
index of expected volatility:
IRS-VI
a
(t. 1) =
_
1
1 t
P
+
var,a
(t. 1) (25)
where P
+
var,a
(t. 1) is the Standardized IRV swap rate in Eq. (10). We label the index in Eq. (25)
IRS-VI.
In Appendix C.1, we show that the expression in Eq. (25) can be simplied and fed with
only the Blacks (1976) volatilities on the out-of-the money swaption skew (see Eq. (C.5)). In
the Appendix, we also show that if \
a
(t. 1) and the path of the short-term rate in the time
interval [t. 1] were uncorrelated, the IRS-VI in Eq. (25) would be the risk-neutral expectation
of the future realized variance dened as:
E-Vol
a
(t. 1) =
_
1
1 t
1
Q
[\
a
(t. 1)]. (26)
22 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
where 1
Q
denotes expectation in a risk-neutral market.
In practice, \
a
(t. 1) and interest rates are likely to be correlated. For example, if the short-
term rate is generated by the Vasicek (1977) model, the correlation between \
a
(t. 1) and
PVBP
T
(1
1
. . 1
a
) is positivea nding we established as a by-product of experiments re-
ported in Appendix F. Despite this theoretical dependence, Eq. (25) is, for all intents and
purposes, the same as the risk-neutral expectation of future realized volatility, Eq. (26). Figure
7 plots the relation between volatility and the forward rate arising within the Vasicek market.
The left panels plot the relation between the forward swap rate and: (i) the IRS-VI
a
(t. 1) index
in Eq. (25), for a maturity 1 t = 1 month and tenor : = 5 years (top panels), a maturity
1 t = 1 year and tenor : = 10 years (bottom panels), and regular quarterly reset dates; and
(ii) the risk-neutral expectation of future realized volatility of the forward swap rate, computed
through Eq. (26). The right panels plot the relation between the forward swap rate and the
instantaneous volatility of the forward swap rate, as dened in Appendix F, Eq. (F.4). Ap-
pendix F provides details regarding these calculations, as well as additional numerical results
relating to a number of alternative combinations of maturities and tenors (see Table A.1). In
these numerical examples, the IRS-VI index provides an upper bound to the expected volatility
in a risk-neutral market, and it highly correlates with it. The IRS-VI and expected volatility re-
spond precisely in the same way to changes in market conditions, as summarized by movements
in the forward swap rate, and deviate quite insignicantly from each other. Interestingly, the
Vasicek market is one where we observe the interest rate counterpart of the leverage eect
observed in equity markets: low interest rates are associated with high interest rate volatility.
7.2. Basis point
The BP counterpart to the IRS-VI index in Eq. (25) is:
IRS-VI
BP
a
(t. 1) =
_
1
1 t
P
+BP
var,a
(t. 1). P
+BP
var,a
(t. 1) =
F
BP
var,a
(t. 1)
PVBP
t
(1
1
. . 1
a
)
(27)
where F
BP
var,a
(t. 1) is the price of the IRV forward agreement in Eq. (14). Similar to the percent-
age index in Eq. (25), the Basis Point index in Eq. (27) can be calculated by feeding Blacks
(1976) formula with the swaption skew (see Eq. (G.5) in Appendix G.1).
Interestingly, our model-free BP volatility index squared and rescaled by
1
Tt
coincides with
the conditional second moment of the forward swap rate, as it turns out by comparing our
formula in Eq. (27) with expressions in Trolle and Schwartz (2011). It is an interesting statistical
property, which complements the asset pricing foundations lied down in Sections 4 and 5,
pertaining to security designs, hedging, replication, and those relating to marking to market
and trading strategies in Section 6. In Section 9, we shall show that our Basis Point index enjoys
the additional interesting property of being resilient to jumps: its value remains the same, and
hence model-free, even in the presence of jumps.
Finally, Figure 8 reports experiments that are the basis point counterparts to those in Figure
7. In these experiments, we compare BP volatility as referenced by our index for ve and ten
year tenors, with that arising in a risk-neutral Vasicek market, as well as the instantaneous
BP volatility of the forward rate, as predicted by the Vasicek model. Once again, the index
and expected volatility are practically the same, a property arising in a number of additional
experiments reported in Appendix G.4.
23 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
4 5 6 7 8 9
18
20
22
24
26
28
30
32
34
36
1m-5Y Forward Swap Rate (in %)
1
m
-
5
Y

I
R
S
-
V
I

a
n
d

E
x
p
e
c
t
e
d

V
o
l
a
t
i
l
i
t
y

(
i
n

%
)
IRS-VI (in %)
Expected Volatility (in %)
4 5 6 7 8 9
18
20
22
24
26
28
30
32
34
36
1m-5Y Forward Swap Rate (in %)
F
o
r
w
a
r
d

S
w
a
p

I
n
s
t
a
n
t
a
n
e
o
u
s

V
o
l
a
t
i
l
i
t
y

(
i
n

%
)
5.5 6 6.5 7 7.5 8
11.5
12
12.5
13
13.5
14
14.5
1Y-10Y Forward Swap Rate (i n %)
1
Y
-
1
0
Y

I
R
S
-
V
I

a
n
d

E
x
p
e
c
t
e
d

V
o
l
a
t
i
l
i
t
y

(
i
n

%
)
IRS-VI (i n %)
Expected Vol ati l i ty (i n %)
5.5 6 6.5 7 7.5 8
9
9.5
10
10.5
11
11.5
12
1Y-10Y Forward Swap Rate (i n %)
F
o
r
w
a
r
d

S
w
a
p

I
n
s
t
a
n
t
a
n
e
o
u
s

V
o
l
a
t
i
l
i
t
y

(
i
n

%
)
Figure 7. Left panels: The IRS-VI index and expected volatility (both in percent-
age) in a risk-neutral Vasicek market, as a function of the level of the forward swap
rate. Right panels: the instantaneous forward swap rate volatility (in percentage) as
a function of the forward swap rate. Top panels relate to a length of the variance
contract equal to 1 month and tenor of 5 years. Bottom panels relate to a length of
the variance contract equal to 1 year and tenor of 10 years.
24 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
4 5 6 7 8 9
148
150
152
154
156
158
160
1m-5Y Forward Swap Rate (in %)
1
m
-
5
Y

I
R
S
-
V
I

a
n
d

E
x
p
e
c
t
e
d

V
o
l
a
t
i
l
i
t
y

(
i
n

B
P
)
IRS-VI (in BP)
Expected Volatility (in BP)
4 5 6 7 8 9
146
148
150
152
154
156
158
1m-5Y Forward Swap Rate (in %)
F
o
r
w
a
r
d

S
w
a
p

I
n
s
t
a
n
t
a
n
e
o
u
s

V
o
l
a
t
i
l
i
t
y

(
i
n

B
P
)
5.5 6 6.5 7 7.5 8
81
82
83
84
85
86
87
88
89
1Y-10Y Forward Swap Rate (in %)
1
Y
-
1
0
Y

I
R
S
-
V
I

a
n
d

E
x
p
e
c
t
e
d

V
o
l
a
t
i
l
i
t
y

(
i
n

B
P
)
IRS-VI (in BP)
Expect ed Volatility (in BP)
5.5 6 6.5 7 7.5 8
66
67
68
69
70
71
72
73
1Y-10Y Forward Swap Rate (in %)
F
o
r
w
a
r
d

S
w
a
p

I
n
s
t
a
n
t
a
n
e
o
u
s

V
o
l
a
t
i
l
i
t
y

(
i
n

B
P
)
Figure 8. Left panels: The IRS-VI index and expected volatility (both in Basis
Points) in a risk-neutral Vasicek market, as a function of the level of the forward
swap rate. Right panels: the instantaneous forward swap rate volatility (in Basis
Points) as a function of the forward swap rate. Top panels relate to a length of the
variance contract equal to 1 month and tenor of 5 years. Bottom panels relate to a
length of the variance contract equal to 1 year and tenor of 10 years.
25 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
8. Rates versus equity variance contracts and indexes
The pricing framework underlying IRV contracts and indexes relies on spanning arguments,
such as those in Bakshi and Madan (2000) and Carr and Madan (2001), as is also the case with
equity variance contracts and indexes. However, IRV contracts and indexes quite dier from
the equity case for a number of reasons.
First, the payos of IRV contracts introduced in this paper have two sources of uncertainty:
one related to the realized variance of the forward swap rate, and another related to the forward
PVBP. These features of the contract design and the assumption that swap rates are stochastic
distinguish our methodology from the work that has been done on pricing equity volatility. In
the equity case, the volatility index is:
VIX(t. 1) =

_
1
1 t
2
c
v(Tt)
_
_

i:1
i
<1t(T)
Put
t
(1
i
. 1)
1
2
i
1
i
+

i:1
i
1t(T)
Call
t
(1
i
. 1)
1
2
i
1
i
_
_
.
(28)
where : is the instantaneous interest rate, assumed to be constant, and Put
t
(1. 1) and
Call
t
(1. 1) are the market prices as of time t of out-of-the money European put and call
equity options with strike prices equal to 1 and time to maturity 1 t, 1
t
(1) is the forward
price of equity, and 1
i
are as in Eq. (8). Note that due to constant discounting, the fair strike
for an equity variance contract is the same as the index, up to time-rescaling and squaring, and
is given by:
P
equity
(t. 1) = (1 t) VIX
2
(t. 1) . (29)
As for IRV contracts, note that the fair value of the Standardized IRV swap rate, P
+
var,a
(t. 1)
in Eq. (7), is still the index, up to time-rescaling and squaring and, by Eq. (25), is given by:
IRS-VI
a
(t. 1)
=

_
1
1 t
2
PVBP
t
(1
1
. . 1
a
)
_

i:1
i
<1t
SWPN
R
t
(1
i
. 1; 1
a
)
1
2
i
1
i
+

i:1
i
1t
SWPN
P
t
(1
i
. 1; 1
a
)
1
2
i
1
i
_
.
(30)
While the two indexes in Eqs. (28) and (30) aggregate prices of out-of-the money derivatives
using the same weights, they dier for three reasons. (i) the VIX in Eq. (28) is constructed
by rescaling a weighted average of out-of-the money option prices through the inverse of the
price of a zero with the same expiry date as that of the variance contract in a market with
constant interest rates, c
v(Tt)
. Instead, the IRS-VI index in Eq. (30) rescales a weighted
average of out-of-the money swaption prices with the inverse of the price of a basket of bonds,
PVBP
t
(1
1
. . 1
a
), each of them expiring at the end-points of the swaps xed payment dates;
(ii) the dierent nature of the derivatives involved in the denition of the two indexes: European
options for the VIX(t. 1) index, and swaptions for the IRS-VI
a
(t. 1) index; (iii) the extra
dimension in swap rate volatility introduced by the length of the tenor period underlying the
forward swap rate, 1
a
1
0
.
There are further distinctions to be made. Consider the current value of a forward for delivery
of the rate variance at time 1 in Section 4.1 (Denition I), which by Eq. (8) is:
F
var,a
(t. 1) = PVBP
t
(1
1
. . 1
a
) P
var,a
(t. 1) . (31)
26 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
As for equity, the price to be paid at t, for delivery of equity volatility at 1, is, instead:
F
equity
(t. 1) = 1
t
(1) P
equity
(t. 1) . (32)
The two prices, F
var,a
and F
equity
, scale up to the time t fair values of their respective notionals
at time 1: (i) the fair value PVBP
t
(1
1
. . 1
a
) of the random notional PVBP
T
(1
1
. . 1
a
),
in Eq. (31); and (ii) the fair value 1
t
(1) of the deterministic notional of one dollar, in Eq.
(32). This remarkable property arises in spite of the dierent assumptions underlying the two
contracts: (i) random interest rates, for the rate variance contract, and (ii) constant interest
rates, for the equity variance contract.
Finally, consider the Basis Point IRV contract in Section 5. By using the expression in Eq.
(14) for the fair value of the Basis Point Standardized IRV swap rate, the BP index in Eq. (27)
can be written as:
IRS-VI
BP
a
(t. 1)
=

_
1
1 t
2
PVBP
t
(1
1
. . 1
a
)
_

i:1
i
<1t
SWPN
R
t
(1
i
. 1; 1
a
) 1
i
+

i:1
i
1t
SWPN
P
t
(1
i
. 1; 1
a
) 1
i
_
.
(33)
For the index of the Basis Point volatility metric in Eq. (33), the weights to be given to the
out-of-the money swaption prices are not inversely proportional to the square of the strike, as
in the case for the equity volatility index in Eq. (28). In the theoretical case where we were
given a continuum of strikes, our BP index would actually be an equal-weighted average of out-
of-the money swaption prices, as revealed by the exact expression of the BP-Standardized IRV
swap rate in Appendix G (see Eq. (G.4)), and consistently with the constant Gaussian vega
explanations of Section 5.1 (see Eq. (17))although due to discretization and truncation, the
weights in Eq. (33), 1
i
, might not be exactly the same. Therefore, the Basis Point interest
rate volatility index and contracts dier from equity, not only due to the aspects pointed out for
the percentage volatility, but also for the particular weighting each swaption price enters into
the denition of the index and contracts. Note that these dierent weightings reect dierent
hedging strategies. As explained in Sections 4 and 5, replicating a variance contract agreed
over a percentage volatility metric requires relying on log-contracts, whereas replicating a
variance contract agreed over the basis point realized variance, requires relying on quadratic
contracts.
9. Resilience to jumps
This section examines the pricing and indexing of expected interest rate swap volatility in
markets where the forward swap rate is a jump-diusion process with stochastic volatility:
d1
t
(1
1
. . 1
a
)
1
t
(1
1
. . 1
a
)
=
_
E
Qswap;
_
c
)n(t)
1
_
j (t)
_
dt
+o
t
(1
1
. . 1
a
) d\
+
(t) +
_
c
)n(t)
1
_
d`
+
(t) . t [t. 1] . (34)
where \
+
(t) is a multidimensional Wiener process dened under the swap probability Q
swap
,
o
t
(1
1
. . 1
a
) is a diusion component, adapted to \
+
(t), `
+
(t) is a Cox process under
27 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
Q
swap
with intensity equal to j (t), and ,
a
(t) is the logarithmic jump size. (See, e.g., Jacod and
Shiryaev (1987, p. 142-146), for a succinct discussion of jump-diusion processes.) By applying
Its lemma for jump-diusion processes to Eq. (34), we have that
d ln 1
t
(1
1
. . 1
a
) = ( ) dt +o
t
(1
1
. . 1
a
) d\
+
(t) +,
a
(t) d`
+
(t) .
such that the realized variance of the logarithmic changes of the forward swap rate over a time
interval [t. 1], or percentage variance, is now:
\
J
a
(t. 1) =
_
T
t
|o
t
(1
1
. . 1
a
)|
2
dt +
_
T
t
,
2
a
(t) d`
+
(t) . (35)
Instead, the realized variance of the arithmetic changes of the forward swap rate over [t. 1],
or basis point variance, is:
\
J,BP
a
(t. 1) =
_
T
t
1
2
t
(1
1
. . 1
a
) |o
t
(1
1
. . 1
a
)|
2
dt+
_
T
t
1
2
t
(1
1
. . 1
a
)
_
c
)n(t)
1
_
2
d`
+
(t) .
(36)
These denitions generalize those in Eq. (5) and Eq. (12).
We now derive the fair value of the IRV Standardized contracts in Sections 4 and 5 in this
new setting. In Appendix H, we show that the fair value of the percentage IRV contract payo
Var-Swap
+
a
(t. 1) in Eq. (7), P
J,+
var,a
(t. 1) say, generalizes that in Eq. (10) because of the presence
of one additional term, capturing jumps, as follows:
P
J,+
var,a
(t. 1) = P
+
var,a
(t. 1) 2E
Qswap;
__
T
t
_
c
)n(t)
1 ,
a
(t)
1
2
,
2
a
(t)
_
d`
+
(t)
_
. (37)
where P
+
var,a
(t. 1) is the fair value of the Standardized IRV contract in Eq. (10). Suppose,
for example, that the distribution of jumps is skewed towards negative values. The fair value
P
J,+
var,a
(t. 1) should then be higher than it would be in the absence of jumps. Assume, in par-
ticular, that the distribution of jumps collapses to a single point,

, < 0 say, and that the jump
intensity equals some positive constant j, in which case the fair value in Eq. (37) collapses to,
P
J,+
var,a
(t. 1) = P
+
var,a
(t. 1) + 2 (1 t) j. (38)
where we have dened the positive constant = (c

)
1

,
1
2

,
2
) 0. The index of
percentage swap volatility is now,
IIS-VI
J
a
(t, T) =
_
IIS-VI
2
a
(t, T)
2
T t
E
Qswap;
_
_
T
t
_
c
)n(t)
1 ,
a
(t)
1
2
,
2
a
(t)
_
d
+
(t)
_
(39)
where IRS-VI
a
(t. 1) is as in Eq. (25). In the special case of the parametric example underlying
the price of the IRV contract of Eq. (38), the previous formula reduces to,
IRS-VI
J
a
(t. 1) =
_
IRS-VI
2
a
(t. 1) + 2 j.
In a notable contrast, for the case of basis point IRV, we prove in Appendix H that the
fair value of the basis point IRV contract payo in Denition IV-(c), denoted P
J,+BP
var,a
(t. 1), is
resilient to jumps, meaning
P
J,+BP
var,a
(t. 1) = P
+BP
var,a
(t. 1) . (40)
28 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
where P
+BP
var,a
(t. 1) is as in Eq. (15). Accordingly, the basis point index of swap volatility, IRS-
VI
BP,J
a
(t. 1) say, is the same as that we derived in the absence of jumps,
IRS-VI
BP,J
a
(t. 1) = IRS-VI
BP
a
(t. 1) (41)
where IRS-VI
BP
a
(t. 1) is as in Eq. (27).
10. Index implementation
This section provides examples of a numerical implementation of the index. Section 10.1 devel-
ops a step-by-step illustration of the index calculations, and Section 10.2 contains examples of
the index behavior around particular dates.
10.1. A numerical example
The following example is a non-limiting illustration of the main steps involved into the con-
struction of the IRS-VI indexes IRS-VI
BP
a
(t. 1) and IRS-VI
a
(t. 1) in Eq. (25) and Eq. (27).
We utilize hypothetical data for implied volatilities expressed in percentage terms for swaptions
maturing in one month and tenor equal to ve years. The rst two columns of Table V reports
strike rates, 1 say, and the skew, i.e. the percentage implied volatilities for each strike rate,
denoted as IV(1) (Percentage Implied Vol). For reference, the third column provides basis
point implied volatilities for each strike rate 1, denoted as IV
BP
(1), and computed as:
IV
BP
(1) = IV(1) 1. (42)
where 1 denotes the current forward swap rate (Basis Point Implied Vol). For example, from
Table V, 1 = 2.7352% and IV(1)[
1=1
= 35.80% and, then, IV
BP
(1)

1=1
= 2.7352%
35.80% = 0.979202%, which is 97.9202 basis points volatility. Basis point volatilities are not
needed to compute the indexes of this paper.
The two IRS-VI indexes in Eq. (25) and Eq. (27) are implemented through Blacks (1976)
formula, as explained in Appendix C.1, Eq. (C.5), and Appendix G.1, Eq. (G.5). First, we plug
the skew IV(1) into the Blacks (1976) formulae,
^
2
_
1. 1. 1
i
; (1 t) IV
2
(1
i
)
_
= 2
_
1. 1. 1
i
; (1 t) IV
2
(1
i
)
_
+1 1. (43)
2 (1. 1. 1; \ ) = 1(d) 1(d
_
\ ). d =
ln
1
1
+
1
2
\
_
\
. (44)
where denotes the cumulative standard normal distribution, and
^
2 () and 2 () are the
Blacks prices, i.e. swaption prices (receivers, for
^
2 (); and payers, for 2 ()), divided by the
PVBP. Second, we use Eqs. (43)-(44), and calculate two indexes in Eq. (25) and Eq. (27):
IRS-VIn (t, T)
= 100

_
2
T t
_
_

i:K
i
<R
t
^
Z
_
1, T, 1
i
; (T t) IV
2
(1
i
)
_
1
2
i
1
i
+

i:K
i
R
t
Z
_
1, T, 1
i
; (T t) IV
2
(1
i
)
_
1
2
i
1
i
_
_
, (45)
and:
IRS-VI
BP
n
(t, T)
29 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
= 100 100

_
2
T t
_
_

i:K
i
<R
t
^
Z
_
1, T, 1
i
; (T t) IV
2
(1
i
)
_
1
i
+

i:K
i
R
t
Z
_
1, T, 1
i
; (T t) IV
2
(1
i
)
_
1
i
_
_
. (46)
The basis point index in Eq. (46), is rescaled by 100
2
, to mimic the market practice to express
basis point implied volatility as the product of rates times log-volatility, where both rates and
log-volatility are multiplied by 100.
The fourth column of Table V provides the values of
^
2 and 2 (Blacks prices), for each
strike rate.
Table V
Blacks prices
Strike
Rate (%)
Percentage
Implied Vol
Basis Point
Implied Vol
Receiver
Swaption (
^
Z)
Payer
Swaption Z
1.7352 36.1900 98.9869 ~ 0 10.000010
3
1.9852 36.1900 98.9869 0.000710
3
7.500710
3
2.2352 36.1200 98.7954 0.025910
3
5.025910
3
2.4352 35.9900 98.4398 0.177310
3
3.177310
3
2.5352 35.9300 98.2757 0.369210
3
2.369210
3
2.6352 35.8600 98.0843 0.679310
3
1.679310
3
2.6852 35.8300 98.0022 0.885510
3
1.385510
3
2.7352 35.8000 97.9202 1.127210
3
1.127210
3
2.7852 35.7600 97.8108 1.403710
3
0.903710
3
2.8352 35.7300 97.7287 1.714210
3
0.714210
3
2.9352 35.6700 97.5646 2.427010
3
0.427010
3
3.0352 35.6000 97.3731 3.240610
3
0.240610
3
3.2352 35.4700 97.0175 5.064410
3
0.064410
3
3.4852 35.3100 96.5799 7.509210
3
0.009210
3
3.7352 35.1400 96.1149 10.001010
3
0.001010
3
Table VI provides details regarding the computation of the indexes IRS-VI
BP
a
(t. 1) and IRS-
VI
a
(t. 1) through Eqs. (46) and (45): the second column displays the type of out-of-the money
swaption entering into the calculation; the third column has the corresponding Blacks prices;
the fourth and fth columns report the weights each price bears towards the nal computation of
the index before the nal rescaling of
2
Tt
; nally, the sixth and seventh columns report each out-
of-the money swaption price multiplied by the appropriate weight (i.e. third column multiplied
by the fourth column for the Basis Point Contribution , and third column multiplied by the
fth column for the Percentage Contribution.
Table VI
Weights Contributions to Strikes
Strike
Rate (%)
Swaption
Type Price
Basis Point
1
i
Percentage
1
i
1
2
i
Basis Point
Contribution
Percentage
Contribution
1.7352 Receiver ~ 0 0.0025 8.3031 ~ 0 ~ 0
1.9852 Receiver 0.000710
3
0.0025 6.3435 0.001810
6
0.004610
3
2.2352 Receiver 0.025910
3
0.0022 4.5035 0.058310
6
0.116710
3
2.4352 Receiver 0.177310
3
0.0015 2.5294 0.266010
6
0.448510
3
2.5352 Receiver 0.369210
3
0.0010 1.5559 0.369210
6
0.574410
3
2.6352 Receiver 0.679310
3
0.0008 1.0800 0.509510
6
0.733710
3
2.6852 Receiver 0.885510
3
0.0005 0.6935 0.442810
6
0.614110
3
2.7352 ATM 1.127210
3
0.0005 0.6683 0.563610
6
0.753310
3
2.7852 Payer 0.903710
3
0.0005 0.6446 0.451810
6
0.582510
3
2.8352 Payer 0.714210
3
0.0007 0.9330 0.535710
6
0.666410
3
2.9352 Payer 0.427010
3
0.0010 1.1607 0.427010
6
0.495610
3
3.0352 Payer 0.240610
3
0.0015 1.6282 0.360910
6
0.391710
3
3.2352 Payer 0.064410
3
0.0023 2.1497 0.144810
6
0.138410
3
3.4852 Payer 0.009210
3
0.0025 2.0582 0.022910
6
0.018810
3
3.7352 Payer 0.001010
3
0.0025 1.7919 0.002410
6
0.001710
3
SUMS 4.156710
6
5.540510
3
The two indexes are computed as
IRS-VI
a
(t. 1) = 100
_
2
12
1
5.5405 10
3
= 36.4653.
30 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
IRS-VI
BP
a
(t. 1) = 100 100
_
2
12
1
4.1567 10
6
= 99.8803.
In comparison, ATM implied volatilities are IV(1) = 35.8000 and IV
BP
(1) = 97.9202.
10.2. Historical performance
This section documents the performance of the IRS-VI index over selected days by relying on
data provided by a major interdelear broker.
First, we consider the index behavior at four selected dates: (i) February 16, 2007, (ii)
February 15, 2008, (iii) February 13, 2009, and (iv) February 12, 2010.
Second, we document the index behavior over the Lehmans collapse occurred in Septem-
ber 2008.
We calculate the index through Eq. (25), using ATM swaptions and out-of-the-money (OTM,
henceforth) swaptions, with moneyness equal to (200, (100, (50, and, nally, (25 basis points
away from the current forward swap rate. We report the index and implied volatilities in per-
centages. Basis point volatilities are percentage implied volatilities multiplied by the current
forward swap rate. The Basis Point IRS-VI index of Section 5, IRS-VI
BP
a
(t. 1), dened in Eq.
(27), aims to aggregate information about these volatilities, in that it tracks expected volatil-
ity cast in basis point terms. We only report experiments relating to the percentage index, as
dened in Eq. (25), IRS-VI
a
(t. 1). Furthermore, note that the index is calculated using rolling
tenors, as for the trading strategy in Figure 9: for each time-to-maturity 1, IRS-VI
a
(t. 1) is
the volatility index for tenor period equal to 1
a
1, for all considered :.
Figures 9 through 12 compare the IRS-VI index with the implied volatility for at-the-money
swaptions with ve year tenors (ATM volatility, henceforth), as a function of the maturity of
the variance contract (one, three, six, nine and twelve months), over the selected dates. Figures
13 through 16 plot the IRS-VI surface for the same dates, that is, the plot of the IRS-VI
index against (i) the maturity of the variance contract, and (ii) the tenor of the forward rate.
Figures 9-12 show that quantitatively, the IRS-VI index behaves dissimilarly from ATM
volatilities. It may be larger or smaller than ATM volatilities, depending on the specic date
at which it is measured or the horizon of the variance contract. For example, on February
15, 2008, the IRS-VI index is higher than ATM volatilities for low maturities of the variance
contract, and is lower otherwise. The reason for these dierences is that the IRS-VI aggregates
information relating to both ATM and OTM volatilities, in such a way to isolate expected
volatility from other factors, such as changes in the underlying forward rates. ATM volatilities,
instead, and by default, only rely on the current forward swap rates. On February 12, 2010,
to cite a second example, OTM volatilities were in general higher (resp. lower) than ATM
volatilities, in correspondence of smaller (larger) maturities. This pattern is captured by the
IRS-VI, which is higher than ATM volatilities for maturities up to six months, and lower than
ATM volatilities for maturities of nine and twelve months.
Qualitatively, the shape of the IRS-VI index displays more pronounced characteristics than
ATM volatilities, when assessed against the length of the variance contract. In two days oc-
curring during bad times (February 15, 2008 and February 13, 2009), the IRS-VI slopes
downwards more sharply than ATM volatilities do, thereby vividly signaling the market expec-
tation that bad times will be followed by periods where resolution of uncertainty about interest
31 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
rates will occur. In good times (February 16, 2007), the IRS-VI index and ATM volatilities
both slope up, although moderately, reecting the market concern these good times, while per-
sistent, will be slightly more uncertain in the future; even in this simple case, the IRS-VI index
exhibits a richer behavior than ATM volatilities, as it attens out for maturities higher than
nine months. Finally, the hump shape of the IRS-VI index and ATM volatilities occurring on
February 12, 2010 reects the market expectation that the uncertainty about interest rates will
be rising in the immediate, although then it will subside in the following months.
Naturally, adding to the fact the IRS-VI index and ATM volatilities are simply not the same
thing, both quantitatively and qualitatively, variance contracts such as those in Section 4 cannot
be priced using solely ATM volatilities. Rather, these contracts need to rely upon the IRS-VI
index, as explained in the previous sections. Figures 13 through 16 show how the IRS-VI index
can be used to express views about developments in interest rate swap volatility. In fact, one of
the added features of the IRS-VI index and contract, compared to the VIX
R _
index for equities,
is the possibility for market participants to choose the tenor of the forward swap rate, on top
of the period length about which to convey volatility views. This added dimension arises quite
naturally, as a result of an increased complexity of xed income derivatives, as opposed to
equity. As an example, the trading strategies of Figure 5 can be implemented on a particular
tenor period 1
a
1, and trading can take place over one of the tenors in Figure 13-16, where
the mispricing is thought to be occurring.
Figures 17 through 21 depict the behavior of the index around the Lehmans collapse, occurred
on September 15, 2008. A few days before the collapse, the index was downward sloping with
respect to maturity, reecting market expectations of possibly imminent negative tail events.
These expectations would reinforce the day of the Lehmans collapse, and only partially weaken
over the week of this event.
32 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
1 month 3 months 6 months 9 months 12 months
12
12.5
13
13.5
14
14.5
15
IRS-VI for 5 year tenor (i n %)
ATM Swapti on Impl i ed Vol ati l i ty (i n %)
Figure 9. Term structure of the Interest Rate Swap Volatility Index (IRS-VI)
February 16, 2007.
1 month 3 months 6 months 9 months 12 months
25
30
35
40
45
IRS-VI f or 5 year tenor (in %)
ATM Swaption Implied Volatility (in %)
Figure 10. Term structure of the Interest Rate Swap Volatility Index (IRS-VI)
February 15, 2008.
33 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
1 month 3 months 6 months 9 months 12 months
35
40
45
50
55
60
65
IRS-VI f or 5 year tenor (in %)
ATM Swaption Implied Volatility (in %)
Figure 11. Term structure of the Interest Rate Swap Volatility Index (IRS-VI)
February 13, 2009.
1 month 3 months 6 months 9 months 12 months
30
31
32
33
34
35
36
37
38
39
40
IRS-VI f or 5 year tenor (in %)
ATM Swaption Implied Volatility (in %)
Figure 12. Term structure of the Interest Rate Swap Volatility Index (IRS-VI)
February 12, 2010.
34 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
2y
3y
4y
5y
6y
7y
8y
9y
10y
1m
3m
6m
9m
12m
12.5
13
13.5
14
14.5
15
15.5
Length of the volatility
contract, in months
Underlying swap tenor, in years
I
R
S
-
V
I

(
i
n

%
)
Figure 13. Interest Rate Swap Volatility Index (IRS-VI)February 16, 2007.
2y
3y
4y
5y
6y
7y
8y
9y
10y
1m
3m
6m
9m
12m
20
25
30
35
40
45
50
55
Length of the volatility
contract, in months
Underlying swap tenor, in years
I
R
S
-
V
I

(
i
n

%
)
Figure 14. Interest Rate Swap Volatility Index (IRS-VI)February 15, 2008.
35 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
2y
3y
4y
5y
6y
7y
8y
9y
10y
1m
3m
6m
9m
12m
30
40
50
60
70
80
Length of the volatility
contract, in months
Underlying swap tenor, in years
I
R
S
-
V
I

(
i
n

%
)
Figure 15. Interest Rate Swap Volatility Index (IRS-VI)February 13, 2009.
2y
3y
4y
5y
6y
7y
8y
9y
10y
1m
3m
6m
9m
12m
20
30
40
50
60
70
Length of the volatility
contract, in months
Underlying swap tenor, in years
I
R
S
-
V
I

(
i
n

%
)
Figure 16. Interest Rate Swap Volatility Index (IRS-VI)February 12, 2010.
36 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
2y
3y
4y
5y
6y
7y
8y
9y
10y
1m
3m
6m
9m
12m
20
25
30
35
40
45
50
Lengt h of t he v olat ilit y
cont ract , in mont hs
Sept ember 10, 2008 (Wednesday )
Underly ing swap t enor, in y ears
I
R
S
-
V
I

(
i
n

%
)
Figure 17. Interest Rate Swap Volatility Index (IRS-VI) around the Lehmans
collapse days: I.
2y
3y
4y
5y
6y
7y
8y
9y
10y
1m
3m
6m
9m
12m
20
25
30
35
40
45
50
Lengt h of t he v olat ilit y
cont ract , in mont hs
Sept ember 12, 2008 (Friday )
Underly ing swap t enor, in y ears
I
R
S
-
V
I

(
i
n

%
)
Figure 18. Interest Rate Swap Volatility Index (IRS-VI) around the Lehmans
collapse days: II.
37 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
2y
3y
4y
5y
6y
7y
8y
9y
10y
1m
3m
6m
9m
12m
20
30
40
50
60
70
80
90
Lengt h of t he v olat ilit y
contract, in months
Sept ember 15, 2008 (Monday , Lehman's collapse)
Underly ing swap tenor, in y ears
I
R
S
-
V
I

(
i
n

%
)
Figure 19. Interest Rate Swap Volatility Index (IRS-VI) around the Lehmans
collapse days: III.
2y
3y
4y
5y
6y
7y
8y
9y
10y
1m
3m
6m
9m
12m
20
30
40
50
60
70
Lengt h of t he v olat ilit y
cont ract , in mont hs
Sept ember 16, 2008 (Tuesday )
Underly ing swap t enor, in y ears
I
R
S
-
V
I

(
i
n

%
)
Figure 20. Interest Rate Swap Volatility Index (IRS-VI) around the Lehmans
collapse days: IV.
38 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
2y
3y
4y
5y
6y
7y
8y
9y
10y
1m
3m
6m
9m
12m
25
30
35
40
45
50
55
60
65
Lengt h of t he v olat ilit y
cont ract , in mont hs
Sept ember 18, 2008 (Thursday )
Underly ing swap t enor, in y ears
I
R
S
-
V
I

(
i
n

%
)
Figure 21. Interest Rate Swap Volatility Index (IRS-VI) around the Lehmans
collapse days: V.
39 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
Technical Appendix
A. Notation, assumptions and preliminary facts
Let 1
t
(T) be the price as of time t of a zero coupon bond expiring at time T t.
The forward swap rate as of time t is the xed interest rate that makes the value of a for-
ward starting swap equal to zero. For a forward start swap contract with future reset dates
T
0
, , T
a1
and payment periods T
1
T
0
, , T
a
T
a1
, it is dened as:
1
t
(T
1
, , T
a
) =
1
t
(T
0
) 1
t
(T
a
)
IVLI
t
(T
1
, , T
a
)
, (A.1)
where IVLI
t
(T
1
, , T
a
) is the price value of the basis point of the swap,
IVLI
t
(T
1
, , T
a
) =
a

i=1
c
i1
1
t
(T
i
) , (A.2)
and c
i
= T
i+1
T
i
are the lengths of the reset intervals.
Let r
c
be the instantaneous interest rate as of time :, and dene the value of the money market
account as of time t _ t (MMA, for short) as '
t
= c
_
t
t
vsoc
.
Let Q be the risk-neutral probability, under which all traded assets discounted by the MMA are
martingales. Dene the Radon-Nikodym derivative of Q
swap
with respect to Q as:
dQ
swap
dQ

F
T
= c

_
T
t
vsoc
IVLI
T
(T
1
, , T
a
)
IVLI
t
(T
1
, , T
a
)
, (A.3)
where F
T
is the information set as of time T, E
Q
[[ denotes the expectation taken under the
risk-neutral probability, and E
Qswap
[[ is the expectation taken under the Q
swap
probability.
The forward swap rate is solution to (see, e.g., Mele, 2011, Chapter 12):
d1
c
(T
1
, , T
a
)
1
c
(T
1
, , T
a
)
= o
c
(T
1
, , T
a
) d\
+
c
, : [t, T[ , (A.4)
where \
+
t
is a multidimensional Wiener process under the Q
swap
probability, and o
t
(T
1
, , T
a
)
is adapted to \
+
t
.
The payo of the IRV forward agreement is:
\
a
(t, T) IVLI
T
(T
1
, , T
a
) , (A.5)
where \
a
(t, T) is the realized variance of the forward swap rate logarithmic changes in the
interval [t, T[,
\
a
(t, T) =
_
T
t
o
2
c
(T
1
, , T
a
) d:, (A.6)
where we use the simplied notation,
o
2
t
(T
1
, , T
a
) = |o
t
(T
1
, , T
a
)|
2
.
40 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
In the absence of arbitrage, the value of a forward starting swap payer with xed interest 1 is
(see, e.g., Mele, 2011, Chapter 12),
SWAI
t
(1; T
1
, , T
a
) = IVLI
t
(T
1
, , T
a
) [1
t
(T
1
, , T
a
) 1[ , (A.7)
and the prices of a payer and receiver swaptions expiring at T are, respectively,
SWIN
P
t
(1, T; T
1
, , T
a
) = IVLI
t
(T
1
, , T
a
) E
Qswap
[1
T
(T
1
, , T
a
) 1[
+
, (A.8)
and
SWIN
R
t
(1, T; T
1
, , T
a
) = IVLI
t
(T
1
, , T
a
) E
Qswap
[1 1
T
(T
1
, , T
a
)[
+
. (A.9)
The parity for payer and receiver swaptions is:
SWIN
P
t
(1, T; T
1
, , T
a
) = SWAI
t
(1; T
1
, , T
a
) SWIN
R
t
(1, T; T
1
, , T
a
) . (A.10)
B. P&L of option-based volatility trading
B.1. Definitions
The price of a payer swaption in Eq. (A.8) is known in closed-form, once we assume that the volatility
o
t
in Eq. (A.4) and, hence, the integrated variance \
a
(t, T) in Eq. (A.6), are deterministic. It is given
by Blacks (1976) formula:
SWIN_Ll
P
t
_
1
a
t
, IVLI
t
, 1, T;

\
_
= IVLI
t
7
t
_
1
a
t
, T, 1;

\
_
, (B.1)
where:
7
t
(1
t
; T, 1;

\ ) = 1
t
1(d
t
) 11(d
t

\ ), d
t
=
ln
1t
1

1
2

\
_

\
,
1 denotes the cumulative standard normal distribution, and to alleviate notation, 1
a
t
= 1
t
(T
1
, , T
a
),
IVLI
t
= IVLI
t
(T
1
, , T
a
), and

\ is the constant value of \
a
(t, T) in Eq. (A.6). By the denition
of the forward swap rate in Eq. (A.1), we have:
SWIN_Ll
P
t
_
1
a
t
, IVLI
t
, 1, T;

\
_
= (1
t
(T) 1
t
(T
a
)) 1(d
t
) IVLI
t
11(d
t

\ ). (B.2)
Eq. (B.2) shows the swaption can be hedged through portfolios of zero coupon bonds: (i) long 1(d
t
)
units of a portfolio which is long one zero expiring at T and short one zero expiring at T
a
, and (ii)
short 11(d
t
:) units of the bonds basket IVLI
t
. Alternatively, note that:
SWIN_Ll
P
t
_
1
a
t
, IVLI
t
, 1, T;

\
_
= SWAI
t
(1; T
1
, , T
a
) 1(d
t
)IVLI
t
1
_
1(d
t
) 1(d
t

\ )
_
.
(B.3)
Eq. (B.3) shows the swaption can equally be hedged as follows: (i) long 1(d
t
) units of a swap, and
(ii) long 1(1(d
t
) 1(d
t

\ )) units of the bonds basket IVLI

t
.
Next, and against the assumption underlying Eq. (B.2) and Eq. (B.3), we assume the forward swap
rate has stochastic volatility, o
t
say, as in Eq. (A.4), which it does, for example, in the models of
Appendix F. We wish to have a view that volatility will raise, say, compared to the current implied
volatility, dened as the value,
IV
t
=
_

\
T t
, (B.4)
such that, once re-normalized again by T t and, then, inserted into Eq. (B.2), it delivers the swaption
market price.
41 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
B.2. Delta-hedged swaptions strategies
Consider, rst, the strategy to purchase the swaption, and hedge it using one of the portfolios under-
lying either Eq. (B.2) or Eq. (B.3). We derive the P&L of this strategy assuming the hedging portfolio
is that relying on Eq. (B.2). The value of the hedging strategy, say, satises:

t
= a
t
(1
t
(T) 1
t
(T
a
)) /
t
IVLI
t
, t [t, T[ ,
where, denoting for simplicity 1
t
= 1
a
t
,

t
= SWIN_Ll
P
t
_
1
t
, IVLI
t
, 1, T, (T t) IV
2
t
_
,
a
t
= 1
_
ln
1
1

1
2
(T t) IV
2
t
_
T tIV
t
_
and /
t
= 11
_
ln
1
1

1
2
(T t) IV
2
t
_
T tIV
t
_
.
(B.5)
Because the hedging strategy is self-nanced, d
t
= a
t
d [1
t
(T) 1
t
(T
a
)[ /
t
dIVLI
t
, and, hence:
d
t
=
_
j
b
t

t

_
j
1
t
j
b
t
_
a
t
(1
t
(T) 1
t
(T
a
))
_
dt
_
o
b
t

t

_
o
1
t
o
b
t
_
a
t
(1
t
(T) 1
t
(T
a
))
_
d\
t
,
(B.6)
where \
t
is a Wiener process under the physical probability and, accordingly, j
b
t
and o
b
t
denote the
drift and instantaneous volatility of
oPVBP
PVBP
, and j
1
t
and o
1
t
denote the drift and instantaneous
volatility of
o(1 (T)1 (Tn))
1 (T)1 (Tn)
.
On the other hand, consider the swaption price in Eq. (B.1) with

\ replaced by (T t) IV
2
t
, as in Eq.
(B.4), i.e. SWIN_Ll
P
t
= SWIN_Ll
P
t
_
1
t
, IVLI
t
, 1, T, (T t) IV
2
t
_
. By Its lemma, the denition
of the forward swap rate in Eq. (A.1), and the partial dierential equation satised by the pricing
function 7
t
_
1, T, 1;

\
_
, this price changes as follows:
dSWIN_Ll
P
t
= IVLI
t
d7
t
7
t
dIVLI
t
d7
t
dIVLI
t
= IVLI
t
_
07
t
0t

1
2
0
2
7
t
01
2
1
2
t
IV
2
t
. .
_
=0
dt IVLI
t
_
1
2
0
2
7
t
01
2
1
2
t
_
o
2
t
IV
2
t
_

07
t
01
j
1
t
1
t
_
dt

_
j
b
t
SWIN_Ll
P
t

07
t
01
o
b
t
o
t
(1
t
(T) 1
t
(T
a
))
_
dt

_
07
t
01
o
t
(1
t
(T) 1
t
(T
a
)) o
b
t
SWIN_Ll
P
t
_
d\
t
, (B.7)
where j
1
t
is the drift of
o1
1
under the physical probability, and equals
j
1
t
= j
j
t
j
b
t
o
t
o
b
t
, o
t
= o
j
t
o
b
t
, (B.8)
where the second relation follows by Its lemma. By using again the denition of the forward swap rate
in Eq. (A.1), and then Eq. (B.8) and the relation a
t
=
0Z
01
, we can integrate the dierence between
dSWIN_Ll
P
t
in Eq. (B.7), and d
t
in Eq. (B.6), so as to obtain that the P&L at the swaption maturity
is:
SWIN_Ll
P
T

T
= [1
T
1[
+

T
=
1
2
_
T
t
0
2
7
t
01
2
1
2
t
_
o
2
t
IV
2
t
_
(A
t,T
IVLI
t
) dt
_
T
t
A
t,T
o
b
t
_
SWIN_Ll
P
t

t
_
d\
t
,
(B.9)
42 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
where we have used the rst relation in Eqs. (B.5), SWIN_Ll
P
t
=
t
, and dened A
t,T
= c
_
T
t
j
b
s
oc
. The
approximation in Eq. (2) relies on: (i) IVLI
T
- A
t,T
IVLI
t
; (ii)
1t
1t
- d\
t
; and (iii) disregarding
the term A
t,T
inside the stochastic integral in Eq. (B.9), which we merely made to simplify the
presentation.
B.3. Straddles strategies
Finally, we consider the P&L relating to trading the straddle. The value of the straddle is, S1IADDLL
t
= SWIN
P
t
SWIN
R
t
. By Blacks (1976) formula, the price of the receiver swaption is:
SWIN_Ll
R
t
_
1
a
t
, IVLI
t
, 1, T;

\
_
= IVLI
t

7
t
_
1
a
t
, T, 1;

\
_
,

7
t
_
1
t
, T, 1;

\
_
= 1
_
1 1(d
t

\ )
_
1
t
(1 1(d
t
)) , d
t
=
ln
1t
1

1
2

\
_

\
.
(B.10)
Therefore, the dynamics of SWIN_Ll
R
t
are the same as SWIN_Ll
P
t
in Eq. (B.7), but with

7
t
replacing 7
t
. Moreover, we have, by Eq. (A.10) and Eq. (A.7),
0S1IADDLL
t
01
= IVLI
t
_
1 2
0

7
t
01
_
,
0

7
t
01
=
07
t
01
1,
where 7
t
is as in Eq. (B.1). Assuming the straddle delta is suciently small, which by the previous
equation it is when 2
0Z
01
- 1, the value of the straddle is, by Eq. (B.7), and the previous arguments
about the dynamics of SWIN_Ll
R
t
:
S1IADDLL
T
S1IADDLL
t
=
_
T
t
0
2
7
t
01
2
1
2
t
_
o
2
t
IV
2
t
_
(A
t,T
IVLI
t
) dt
_
T
t
A
t,T
o
b
t
S1IADDLL
t
d\
t
. (B.11)
The approximation in Eq. (3) relies on the same arguments leading to Eq. (2). Note that the ap-
proximation 2
0Z
01
- 1 is quite inadequate as the forward swap rates drifts away from ATMa very
well-known feature discussed, e.g. by Mele (2011, Chapter 10) in the case of equity straddles.
C. Spanning the contracts
C.1. Pricing
We rely on spanning arguments similar to those utilized by Bakshi and Madan (2000) and Carr and
Madan (2001) for equities. To alleviate notation, we set 1
t
= 1
a
t
= 1
t
(T
1
, , T
a
). By the usual
Taylors expansion with remainder,
ln
1
T
1
t
=
1
1
t
(1
T
1
t
)
_
1t
0
(1 1
T
)
+
1
1
2
d1
_
o
1t
(1
T
1)
+
1
1
2
d1. (C.1)
Multiplying both sides of the previous equation by IVLI
t
(T
1
, , T
a
), and taking expectations under
Q
swap
, dened through Eq. (A.3):
IVLI
t
(T
1
, , T
a
) E
Qswap
_
ln
1
T
1
t
_
=
_
1t
0
SWIN
R
t
(1, T; T
a
)
1
2
d1
_
o
1t
SWIN
P
t
(1, T; T
a
)
1
2
d1,
(C.2)
where we have used (i) the fact that by Eq. (A.4), the forward swap rate is a martingale under Q
swap
,
and (ii) the pricing Equations (A.8) and (A.9).
43 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
By Eq. (A.4), and a change of measure obtained through Eq. (A.3),
2E
Qswap
_
ln
1
T
1
t
_
= E
Qswap
__
T
t
o
2
c
(T
1
, , T
a
) d:
_
=
1
IVLI
t
(T
1
, , T
a
)
E
Q
_
c

_
T
t
vsoc
_
IVLI
T
(T
1
, , T
a
)
_
T
t
o
2
c
(T
1
, , T
a
) d:
__
.
(C.3)
Combining this relation with Eq. (C.2) leaves:
F
var,a
(t, T) = E
Q
_
c

_
T
t
vsoc
_
IVLI
T
(T
1
, , T
a
)
_
T
t
o
2
c
(T
1
, , T
a
) d:
__
= 2
__
1t
0
SWIN
R
t
(1, T; T
1
, , T
a
)
1
2
d1
_
o
1t
SWIN
P
t
(1, T; T
1
, , T
a
)
1
2
d1
_
. (C.4)
By Eqs. (A.5) and (A.6), the left hand side of the previous equation is the price of the IR forward
variance agreement, and Eq. (8) is its approximation. The claim in the main text that if \
a
(t, T) and
the path of the short-term rate in the time interval [t, T[ were independent, the index IRS-VI
a
(t, T)
in Eq. (25) would be the risk-neutral expectation of \
a
(t, T), follows by Eq. (C.3) and the denition
of the Radon-Nikodym derivative of Q
swap
with respect to Q in Eq. (A.3).
To derive the IRV swap rate P
var,a
(t, T) in Eq. (9), note that by Eq. (6), it solves:
0 = E
Q
_
c

_
T
t
vsoc
(\
a
(t, T) IVLI
T
(T
1
, , T
a
) P
var,a
(t, T))
_
,
which, after rearranging terms, yields Eq. (9). As for the Standardized IRV swap rate P
+
var,a
(t, T) in
Eq. (10), note that by Eq. (7), it satises:
0 = E
Q
_
c

_
T
t
vsoc
_
\
a
(t, T) P
+
var,a
(t, T)
_
IVLI
T
(T
1
, , T
a
)
_
,
which by the denition of the Radon-Nikodym derivative in Eq. (A.3), yields Eq. (10), after rearranging
terms.
Finally, as claimed in the main text, Eq. (25) can be simplied, using the Blacks (1976) formula.
By replacing Eq. (8) into Eq. (25), and using the Blacks formulae in Eqs. (B.1) and (B.10),
IIS-VI
a
(t, T)
=

_
2
T t
_
_

i:1
i
<1t

7
t
_
1
a
t
, T, 1
i
; (T t) IV
2
i,t
_
1
2
i
^1
i


i:1
i
1t
7
t
_
1
a
t
, T, 1
i
; (T t) IV
2
i,t
_
1
2
i
^1
i
_
_
,
(C.5)
where the expressions for 7
t
and

7
t
are given in Eqs. (B.1) and (B.10), 1
a
t
is the current forward swap
rate for maturity T and tenor length T
a
T, 1
a
t
= 1
t
(T
1
, , T
a
), and, nally, IV
i,t
denotes the time
t implied percentage volatility for swaptions with strike equal to 1
i
.
C.2. Marking to market
We rst derive Eq. (19). For a given t (t, T), we need to derive the following conditional expectation
of the payo Vai-Swap
a
(t, T) in Eq. (6):
E
t
Q
_
c

_
T
t
vuo&
Vai-Swap
a
(t, T)
_
44 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
= E
t
Q
_
c

_
T
t
vuo&
(\
a
(t, t) \
a
(t, T)) IVLI
T
(T
1
, , T
a
)
_
1
t
(T) P
var,a
(t, T)
= \
a
(t, t) IVLI
t
(T
1
, , T
a
) F
var,a
(t, T)
1
t
(T)
1
t
(T)
F
var,a
(t, T) , (C.6)
where E
t
Q
denotes expectation under Q, conditional upon all information up to time t, and we have used
the denition of the Radon-Nikodym derivative of Q
swap
in Eq. (A.3) and expression for F
var,a
(, T)
in Eq. (C.4). Eq. (19) follows after plugging the expression for P
var,a
(t, T) in Eq. (9) into Eq. (C.6).
Next, we derive Eq. (20). We have, utilizing the expression of Vai-Swap
+
a
(t, T) in Eq. (7),
E
t
Q
_
c

_
T
t
vuo&
Vai-Swap
+
a
(t, T)
_
= E
t
Q
_
c

_
T
t
vuo&
_
\
a
(t, t) \
a
(t, T) P
+
var,a
(t, T)

IVLI
T
(T
1
, , T
a
)
_
= IVLI
t
(T
1
, , T
a
)
_
\
a
(t, t) P
+
var,a
(t, T) P
+
var,a
(t, T)

,
where the second equality follows by the expression for F
var,a
(, T) in Eq. (C.4), and by Eq. (10).
C.3. Hedging
We provide the details leading to Table I, as those for Table II are nearly identical. By Its lemma:
IVLI
T
(T
1
, , T
a
)
_
T
t
o
2
c
(T
1
, , T
a
) d:
= 2IVLI
T
(T
1
, , T
a
)
_
T
t
d1
c
(T
1
, , T
a
)
1
c
(T
1
, , T
a
)
2IVLI
T
(T
1
, , T
a
) ln
1
T
(T
1
, , T
a
)
1
t
(T
1
, , T
a
)
, (C.7)
where, by Eq. (C.1), the expression for the swap value in Eq. (1) and that for the swaption premium
in Eqs. (A.9)-(A.10), the second term on the right hand side is the payo at T of a portfolio set
up at t, which is two times: (a) short 1,1
t
(T
1
, , T
a
) units a xed interest payer swap struck at
1
t
(T
1
, , T
a
), and (b) long a continuum of out-of-the-money swaptions with weights 1
2
d1. This
portfolio is the static position (ii) in Table I of the main text. By Eqs. (C.1), (C.2) and (C.4), its cost
is F
var,a
(t, T). We borrow F
var,a
(t, T) at time t, and repay it back at time T, as in row (iii) of Table
I of the main text.
Next, we derive the self-nancing portfolio of zero coupon bonds (i) in Table I, by designing it to
be worthless at time t, and to replicate the rst term on the right hand side of Eq. (C.7). First, note
that by Eq. (A.1), the forward swap rate as of time : can be replicated through a portfolio that is
long zeros maturing at T
0
and short zeros maturing at T
a
, with equal weights 1,IVLI
c
(T
1
, , T
a
).
Consider, then, a self-nanced strategy investing in (a) this portfolio, which we call the forward swap
rate for simplicity, and (b) a MMA, as dened in Appendix A, with total value equal to:

t
= 0
t
1
t
(T
1
, , T
a
) c
t
'
t
,
where 0
t
are the units in the forward swap rate and c
t
are the units in the MMA. Consider the
following portfolio:

0
t
1
t
(T
1
, , T
a
) = IVLI
t
(T
1
, , T
a
) ,

c
t
'
t
= IVLI
t
(T
1
, , T
a
)
__
t
t
d1
c
(T
1
, , T
a
)
1
c
(T
1
, , T
a
)
1
_
.
(C.8)
We have that
t
=

0
t
1
t


c
t
'
t
satises:

t
= IVLI
t
(T
1
, , T
a
)
_
t
t
d1
c
(T
1
, , T
a
)
1
c
(T
1
, , T
a
)
, (C.9)
45 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
such that

t
= 0, and
T
= IVLI
T
(T
1
, , T
a
)
_
T
t
d1
c
(T
1
, , T
a
)
1
c
(T
1
, , T
a
)
.
Therefore, by going long two portfolios (

0
t
,

c
t
), the rst term on the right hand side and, then, by
the previous results, the whole of the right hand side of Eq. (C.7), can be replicated and, so, hedged,
provided (

0
t
,

c
t
) is self-nanced. We are only left to show (

0
t
,

c
t
) is self-nanced. We have:
d
t
=

0
t
1
t
(T
1
, , T
a
)
d1
t
(T
1
, , T
a
)
1
t
(T
1
, , T
a
)


c
t
'
t
d'
t
'
t
= IVLI
t
(T
1
, , T
a
)
_
d1
t
(T
1
, , T
a
)
1
t
(T
1
, , T
a
)
r
t
dt
_

_
IVLI
t
(T
1
, , T
a
)
_
t
t
d1
c
(T
1
, , T
a
)
1
c
(T
1
, , T
a
)
_
r
t
dt
= IVLI
t
(T
1
, , T
a
)
_
d1
t
(T
1
, , T
a
)
1
t
(T
1
, , T
a
)
r
t
dt
_
r
t

t
dt
=

0
t
1
t
_
d1
t
(T
1
, , T
a
)
1
t
(T
1
, , T
a
)
r
t
dt
_
r
t

t
dt, (C.10)
where the second line follows by the portfolio expressions for (

0
t
,

c
t
) in Eq. (C.8), the third line holds
by Eq. (C.9), and the fourth follows, again, by the expression for

0
t
1
t
in Eq. (C.8). It is easy to check
that the dynamics of
t
in Eq. (C.10) are those of a self-nanced strategy.
D. Directional volatility trades
The P&L displayed in Figure 4 of Section 3 are calculated using daily data from January 1998 to
December 2009, and comprise yield curve data as well as implied volatilities for swaptions. Yield
curve data are needed to compute forward swap rates for a xed 5 year tenor and the related realized
volatilities, and implied volatilities are needed to compute P&Ls, as explained next.
As for the straddle, the strategy is to go long an at-the-money swaption straddle. Let t denote the
beginning of the holding period (one month or three months). The terminal P&L of the straddle, that
of as of time T say, is:
IVLI
T
(T
1
, , T
a
)
_
[1
T
(T
1
, , T
a
) 1[
+
[1 1
T
(T
1
, , T
a
)[
+
_
Sliaoolo
t
(T
1
, , T
a
) ,
where Sliaoolo
t
(T
1
, , T
a
) is the cost of the straddle at t, and T t is either one month (as in the
top panel of Figure 4) or three months (as in the bottom panel of Figure 4).
Instead, the terminal P&L of the volatility swap contract is calculated consistently with Denition
II and Eq. (9), as:
1
T t
_
IVLI
T
(T
1
, , T
a
) \
a
(t, T)
F
+
var,a
(t, T)
1
t
(T)
_
,
where F
+
var,a
(t, T) approximates the unnormalized strike of the contract to be entered at t. Its exact
value, F
var,a
(t, T), is that in Eq. (8). By Eq. (25),
1
T t
F
var,a
(t, T) = IVLI
t
(T
1
, , T
a
) IIS-VI
a
(t, T)
2
,
which we approximate through:
1
T t
F
+
var,a
(t, T) = IVLI
t
(T
1
, , T
a
) A1M
a
(t, T)
2
,
where A1M
a
(t, T) is the at-the-money implied volatility.
46 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
Finally, the realized variance rescaled by T t,
1
Tt
\
a
(t, T), is calculated as:
IV
n
t
=
_
21 :
22
_
1
IV
n
t
, where IV
n
t
=
t

i=t21n
_
ln
1
i
(T
1
, , T
a
)
1
i1
(T
1
, , T
a
)
_
2
,
with : 1, 8, T
1
t =
1
4
and T
a
t = . The variance risk-premium is dened as IV
n
t

A1M
2
a
(t 21 :, T), where T t is either
1
12
(for one month) or
3
12
(for three months). Finally, to
make the P&Ls of the volatility swap contract and the straddle line up to the same order of magnitude,
the pictures in Figure 4 are obtained by re-multiplying the P&Ls of the volatility swap contracts by
1
12
(top panel) and
3
12
(bottom panel).
E. Local volatility surfaces
We develop arguments that hinge upon those Dumas (1995) and Britten-Jones and Neuberger (2000)
put forth in the equity case. Consider the price of the payer swaption in Eq. (A.8), SWIN
P
t
(1, T; T
a
).
For simplicity, let 1
t
= 1
t
(T
1
, , T
a
). We have:
0SWIN
P
t
0T
=
0 lnIVLI
t
0T
SWIN
P
t
IVLI
t

dE
Qswap
(1
T
1)
+
dT
. (E.1)
Since the forward swap rate is a martingale under Q
swap
, with volatility as in Eq. (A.4), we have:
dE
Qswap
(1
T
1)
+
dT
=
1
2
E
Qswap
_
c (1
T
1) 1
2
T
o
2
T

, (E.2)
where c () is the Diracs delta. We can elaborate the right hand side of Eq. (E.2), obtaining:
E
Qswap
_
c (1
T
1) 1
2
T
o
2
T

=
__
c (1
T
1) 1
2
T
o
2
T
c
c
T
(o
T
[ 1
T
) c
m
T
(1
T
)
. .
= joint density of (o
T
,1
T
)
d1
T
do
T
= 1
2
c
m
T
(1) E
Qswap
_
o
2
T

1
T
= 1

= 1
2
0
2
SWPN
P
t
01
2
IVLI
t
E
Qswap
_
o
2
T

1
T
= 1

,
where c
c
T
(o
T
[ 1
T
) denotes the conditional density of o
T
given 1
T
under Q
swap
, c
m
T
(1
T
) denotes the
marginal density of 1
T
under Q
swap
, and the third line follows by a well-known property of option-like
prices. By replacing this result into Eq. (E.2) and then into Eq. (E.1), we obtain:
0SWIN
P
t
0T
=
0 lnIVLI
t
0T
SWIN
P
t

1
2
1
2
0
2
SWIN
P
t
01
2
E
Qswap
_
o
2
T

1
T
= 1

.
Rearranging this equation, we obtain:
o
2
loc
(1, T) = E
Qswap
_
o
2
T

1
T
= 1

= 2
0SWPN
P
t
0T

0 ln PVBPt
0T
SWIN
P
t
1
2
0
2
SWPN
P
t
01
2
. (E.3)
Next, let the volatility of the forward swap rate in Eq. (A.4) be given by:
o
c
= o (1
c
, :)
c
,
where
c
is another random process. Dene:
o (1, :) =
o
loc
(1, :)
_
E
Qswap
[
2
c
[ 1
c
= 1[
, (E.4)
47 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
where o
loc
(1, :) is as in Eq. (E.3). Consider the following model of the forward swap rate, under the
Q
swap
probability,
d1
c
1
c
= o (1
c
, :)
c
d\
+
c
, : [t, T[ ,
where o (1
c
, :) is as in Eq. (E.4). This model is, theoretically, capable to match the cross-section of
swaptions without errors, and can be used to price all of the non-traded swaptions in Eq. (8), through
Montecarlo integration.
F. The contract and index in the Vasicek market
We consider two alternative models. In the rst model, Vasiceks (1977), the short-term rate follows a
Gaussian process:
dr
t
= c( r r
t
) dt o

d

\
t
, (F.1)
where

\
t
is a Wiener process under the risk-neutral probability, and c, r and o

are three positive

constants. The price of a zero predicted by this model is 1
t
(r
t
; T) = (T t) c
1(Tt)vt
for two
functions and 1, with obvious notation. The forward swap rate predicted by this model is:
1
t
(r
t
; T
1
, , T
a
) =
1
t
(r
t
; T
0
) 1
t
(r
t
; T
a
)
IVLI
t
(r
t
; T
1
, , T
a
)
, (F.3)
the price value of the basis point is,
IVLI
t
(r
t
; T
1
, , T
a
) =
a

i=1
c
i1
1
t
(r
t
; T
i
) ,
the instantaneous volatility of bond returns is,
Vol-I
t
(r
t
; T) = 1(T t) o

, (F.3)
and, nally, the forward swap rate volatility is,
o
t
(r
t
; T
1
, , T
a
)
=
Vol-I
t
(r
t
; T) 1
t
(r
t
; T) Vol-I
t
(r
t
; T
a
) 1
t
(r
t
; T
a
)
1
t
(r
t
; T) 1
t
(r
t
; T
a
)

a
i=1
c
i1
1
t
(r
t
; T
i
) Vol-I
t
(r
t
; T
i
)
IVLI
t
(r
t
; T
1
, , T
a
)
. (F.4)
We simulate Eq. (F.1) using a Milstein approximation method, for initial values of the short-term
rate in the interval [0.01, 0.10[. We use parameter values set equal to c = 0.8807, r = 0.072 and
o

=
_
0.02 0.2841, taken from Veronesi (2010, Chapter 15), and generate simulated values of
o
t
(r
t
; T
1
, , T
a
), by plugging in the simulated values of r
t
. We use the same parameter values for
the Vasicek model in Eq. (F.1), with o

=
_
0.02o
v
. The two expectations,
E
Q
_
c

_
T
t
vsoc
_
IVLI
T
(r
T
; T
1
, , T
a
)
_
T
t
o
2
c
(r
c
; T
1
, , T
a
) d:
__
and E
Q
__
T
t
o
2
c
(r
c
; T
1
, , T
a
) d:
_
, (F.5)
are estimated through Montecarlo integration. The forward rates in Figure 7 and Table A.1 are ob-
tained by plugging the initial values of the short-term rate into Eq. (F.3).
48 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
T
a
b
l
e
A
.
1
T
h
e
I
R
S
-
V
I
a
(
t
,
T
)
i
n
d
e
x
i
n
E
q
.
(
2
5
)
(
l
a
b
e
l
e
d
I
R
S
-
V
I
)
,
a
n
d
f
u
t
u
r
e
e
x
p
e
c
t
e
d
v
o
l
a
t
i
l
i
t
y
i
n
a
r
i
s
k
-
n
e
u
t
r
a
l
m
a
r
k
e
t
,
_
1
T

t
1
Q
[
\
a
(
t
,
T
)
[
,
w
h
e
r
e
\
a
(
t
,
T
)
i
s
a
s
i
n
E
q
.
(
A
.
6
)
(
l
a
b
e
l
e
d
E
-
V
o
l
)
,
f
o
r
(
i
)
t
i
m
e
-
t
o
m
a
t
u
r
i
t
y
T

t
e
q
u
a
l
t
o
1
,
6
a
n
d
9
m
o
n
t
h
s
,
a
n
d
1
,
2
a
n
d
3
y
e
a
r
s
,
a
n
d
(
i
i
)
t
e
n
o
r
l
e
n
g
t
h
:
e
q
u
a
l
t
o
1
,
2
,
3
a
n
d
5
y
e
a
r
s
,
a
n
d
f
o
r
t
i
m
e
-
t
o
m
a
t
u
r
i
t
y
T

t
e
q
u
a
l
t
o
1
,
6
a
n
d
1
2
m
o
n
t
h
s
,
a
n
d
(
i
i
)
t
e
n
o
r
l
e
n
g
t
h
:
e
q
u
a
l
t
o
1
0
,
2
0
a
n
d
3
0
y
e
a
r
s
.
T
h
e
i
n
d
e
x
a
n
d
e
x
p
e
c
t
e
d
v
o
l
a
t
i
l
i
t
i
e
s
a
r
e
c
o
m
p
u
t
e
d
u
n
d
e
r
t
h
e
a
s
s
u
m
p
t
i
o
n
t
h
e
s
h
o
r
t
-
t
e
r
m
i
n
t
e
r
e
s
t
r
a
t
e
i
s
a
s
i
n
t
h
e
V
a
s
i
c
e
k
m
o
d
e
l
o
f
E
q
.
(
F
.
1
)
,
w
i
t
h
p
a
r
a
m
e
t
e
r
v
a
l
u
e
s
c
=
0
.
8
8
0
7
,
r
=
0
.
0
7
2
a
n
d
o

=
0
.
0
8
8
1
.
M
a
t
u
r
i
t
y
o
f
t
h
e
v
a
r
i
a
n
c
e
c
o
n
t
r
a
c
t
T
e
n
o
r
=
1
y
e
a
r
1
m
o
n
t
h
6
m
o
n
t
h
s
9
m
o
n
t
h
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
1
.
0
0
2
.
1
8
1
2
4
.
8
3
1
2
4
.
8
2
2
.
4
8
1
0
6
.
5
3
1
0
6
.
5
3
3
.
2
7
7
3
.
5
0
7
3
.
5
2
3
.
0
0
3
.
8
0
7
1
.
9
7
7
1
.
9
7
4
.
0
1
6
6
.
3
9
6
6
.
3
8
4
.
5
3
5
3
.
3
9
5
3
.
4
0
5
.
0
0
5
.
4
4
5
0
.
6
2
5
0
.
6
2
5
.
5
4
4
8
.
2
5
4
8
.
2
5
5
.
8
0
4
1
.
9
3
4
1
.
9
4
7
.
0
0
7
.
0
8
3
9
.
0
8
3
9
.
0
8
7
.
0
8
3
7
.
9
3
3
7
.
9
3
7
.
0
8
3
4
.
5
4
3
4
.
5
5
9
.
0
0
8
.
7
3
3
1
.
8
5
3
1
.
8
6
8
.
6
3
3
1
.
2
7
3
1
.
2
7
8
.
3
6
2
9
.
3
8
2
9
.
3
8
1
0
.
0
0
9
.
5
6
2
9
.
1
7
2
9
.
1
7
9
.
4
1
2
8
.
7
8
2
8
.
7
5
9
.
0
0
2
7
.
3
4
2
7
.
3
5
T
e
n
o
r
=
2
y
e
a
r
s
1
m
o
n
t
h
6
m
o
n
t
h
s
9
m
o
n
t
h
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
1
.
0
0
2
.
9
3
7
8
.
1
2
7
8
.
1
2
3
.
5
3
6
0
.
1
1
6
0
.
1
2
3
.
8
4
5
2
.
7
4
5
2
.
7
6
3
.
0
0
4
.
3
1
5
3
.
6
6
5
3
.
6
6
4
.
7
0
4
5
.
4
9
4
5
.
5
0
4
.
9
1
4
1
.
5
9
4
1
.
6
1
5
.
0
0
5
.
6
8
4
0
.
9
3
4
0
.
9
3
5
.
8
8
3
6
.
6
2
3
6
.
6
2
5
.
9
8
3
4
.
3
6
3
4
.
3
7
7
.
0
0
7
.
0
7
3
3
.
1
2
3
3
.
1
2
7
.
0
7
3
0
.
6
6
3
0
.
6
7
7
.
0
6
2
9
.
2
9
2
9
.
3
0
9
.
0
0
8
.
4
7
2
7
.
8
4
2
7
.
8
4
8
.
2
6
2
6
.
3
9
2
6
.
4
0
8
.
1
5
2
5
.
5
3
2
5
.
5
5
1
0
.
0
0
9
.
1
8
2
5
.
8
0
2
5
.
8
0
8
.
8
6
2
4
.
6
8
2
4
.
6
9
8
.
6
9
2
4
.
0
1
2
4
.
0
2
T
e
n
o
r
=
3
y
e
a
r
s
1
m
o
n
t
h
6
m
o
n
t
h
s
9
m
o
n
t
h
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
1
.
0
0
3
.
5
1
5
5
.
7
6
5
5
.
7
6
4
.
0
2
4
5
.
1
8
4
5
.
2
0
4
.
2
8
4
0
.
5
2
4
0
.
5
4
3
.
0
0
4
.
6
8
4
2
.
2
3
4
2
.
2
3
5
.
0
2
3
6
.
5
0
3
6
.
5
1
5
.
2
0
3
3
.
6
9
3
3
.
7
1
5
.
0
0
5
.
8
7
3
4
.
0
4
3
4
.
0
4
6
.
0
4
3
0
.
6
4
3
0
.
6
5
6
.
1
2
2
8
.
8
5
2
8
.
8
7
7
.
0
0
7
.
0
6
2
8
.
5
5
2
8
.
5
5
7
.
0
6
2
6
.
4
4
2
6
.
4
5
7
.
0
5
2
5
.
2
5
2
5
.
2
7
9
.
0
0
8
.
2
7
2
4
.
6
2
2
4
.
6
2
8
.
0
8
2
3
.
2
6
2
3
.
2
7
7
.
9
8
2
2
.
4
7
2
2
.
4
9
1
0
.
0
0
8
.
8
7
2
3
.
0
4
2
3
.
0
4
8
.
6
0
2
1
.
9
5
2
1
.
9
6
8
.
4
5
2
1
.
3
0
2
1
.
3
2
T
e
n
o
r
=
5
y
e
a
r
s
1
m
o
n
t
h
6
m
o
n
t
h
s
9
m
o
n
t
h
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
1
.
0
0
4
.
3
1
3
4
.
6
3
3
4
.
6
2
4
.
6
9
2
9
.
5
6
2
9
.
5
8
4
.
8
9
2
7
.
1
6
2
7
.
1
8
3
.
0
0
5
.
2
0
2
9
.
1
1
2
9
.
1
1
5
.
4
6
2
5
.
7
5
2
5
.
7
7
5
.
5
9
2
4
.
0
5
2
4
.
0
7
5
.
0
0
6
.
1
1
2
5
.
1
7
2
5
.
1
7
6
.
2
3
2
2
.
8
4
2
2
.
8
6
6
.
3
0
2
1
.
6
1
2
1
.
6
3
7
.
0
0
7
.
0
4
2
2
.
2
0
2
2
.
2
0
7
.
0
2
2
0
.
5
5
2
0
.
5
6
7
.
0
1
1
9
.
6
4
1
9
.
6
6
9
.
0
0
7
.
9
7
1
9
.
8
9
1
9
.
8
9
7
.
8
2
1
8
.
7
0
1
8
.
7
1
7
.
7
4
1
8
.
0
1
1
8
.
0
3
1
0
.
0
0
8
.
4
4
1
8
.
9
2
1
8
.
9
2
8
.
2
2
1
7
.
9
0
1
7
.
9
1
8
.
1
1
1
7
.
3
0
1
7
.
3
2
49 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
T
a
b
l
e
A
.
1
c
o
n
t
i
n
u
e
d
M
a
t
u
r
i
t
y
o
f
t
h
e
v
a
r
i
a
n
c
e
c
o
n
t
r
a
c
t
T
e
n
o
r
=
1
y
e
a
r
1
y
e
a
r
2
y
e
a
r
s
3
y
e
a
r
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
1
.
0
0
3
.
6
1
6
4
.
2
5
6
4
.
2
6
4
.
6
7
4
1
.
8
7
4
1
.
9
2
5
.
3
8
3
1
.
6
6
3
1
.
7
2
3
.
0
0
4
.
7
6
4
9
.
1
0
4
9
.
1
0
5
.
4
5
3
6
.
0
6
3
6
.
1
0
5
.
9
2
2
8
.
9
0
2
8
.
9
6
5
.
0
0
5
.
9
1
3
9
.
7
2
3
9
.
7
2
6
.
2
4
3
1
.
6
6
3
1
.
6
9
6
.
4
6
2
6
.
5
8
2
6
.
6
3
7
.
0
0
7
.
0
7
3
3
.
3
5
3
3
.
3
5
7
.
0
4
2
8
.
2
2
2
8
.
2
5
7
.
0
0
2
4
.
6
1
2
4
.
6
5
9
.
0
0
8
.
2
3
2
8
.
7
5
2
8
.
7
5
7
.
8
3
2
5
.
4
6
2
5
.
4
8
7
.
5
4
2
2
.
9
1
2
2
.
9
5
1
0
.
0
0
8
.
8
2
2
6
.
9
0
2
6
.
9
0
8
.
2
3
2
4
.
2
7
2
4
.
2
9
7
.
8
1
2
2
.
1
5
2
2
.
1
9
T
e
n
o
r
=
2
y
e
a
r
s
1
y
e
a
r
2
y
e
a
r
s
3
y
e
a
r
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
1
.
0
0
4
.
1
3
4
7
.
3
0
4
7
.
3
2
5
.
0
1
3
2
.
9
5
3
3
.
0
1
5
.
6
1
2
5
.
6
9
2
5
.
7
6
3
.
0
0
5
.
1
0
3
8
.
6
1
3
8
.
6
2
5
.
6
8
2
9
.
2
7
2
9
.
3
2
6
.
0
7
2
3
.
8
6
2
3
.
9
2
5
.
0
0
6
.
0
7
3
2
.
6
2
3
2
.
6
2
6
.
3
5
2
6
.
3
4
2
6
.
3
8
6
.
5
2
2
2
.
2
7
2
2
.
3
3
7
.
0
0
7
.
0
5
2
8
.
2
5
2
8
.
2
5
7
.
0
2
2
3
.
9
4
2
3
.
9
7
6
.
9
8
2
0
.
8
9
2
0
.
9
4
9
.
0
0
8
.
0
4
2
4
.
9
2
2
4
.
9
2
7
.
6
9
2
1
.
9
4
2
1
.
9
7
7
.
4
4
1
9
.
6
6
1
9
.
7
1
1
0
.
0
0
8
.
5
3
2
3
.
5
3
2
3
.
5
4
8
.
0
3
2
1
.
0
7
2
1
.
1
0
7
.
6
7
1
9
.
1
1
1
9
.
1
5
T
e
n
o
r
=
3
y
e
a
r
s
1
y
e
a
r
2
y
e
a
r
s
3
y
e
a
r
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
1
.
0
0
4
.
5
2
3
6
.
9
2
3
6
.
9
4
5
.
2
8
2
6
.
8
7
2
6
.
9
3
5
.
7
9
2
1
.
4
1
2
1
.
4
8
3
.
0
0
5
.
3
6
3
1
.
4
7
3
1
.
4
9
5
.
8
5
2
4
.
4
1
2
4
.
4
6
6
.
1
8
2
0
.
1
4
2
0
.
2
1
5
.
0
0
6
.
1
9
2
7
.
4
4
2
7
.
4
5
6
.
4
2
2
2
.
3
7
2
2
.
4
1
6
.
5
7
1
9
.
0
2
1
9
.
0
8
7
.
0
0
7
.
0
4
2
4
.
3
3
2
4
.
3
4
7
.
0
0
2
0
.
6
4
2
0
.
6
9
6
.
9
7
1
8
.
0
2
1
8
.
0
8
9
.
0
0
7
.
8
9
2
1
.
8
7
2
1
.
8
8
7
.
5
8
1
9
.
1
7
1
9
.
2
1
7
.
3
6
1
7
.
1
2
1
7
.
1
7
1
0
.
0
0
8
.
3
1
2
0
.
8
2
2
0
.
8
3
7
.
8
7
1
8
.
5
1
1
8
.
5
5
7
.
5
6
1
6
.
7
0
1
6
.
7
6
T
e
n
o
r
=
5
y
e
a
r
s
1
y
e
a
r
2
y
e
a
r
s
3
y
e
a
r
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
1
.
0
0
5
.
0
7
2
5
.
1
9
2
5
.
2
2
5
.
6
5
1
9
.
3
5
1
9
.
4
1
6
.
0
4
1
5
.
8
6
1
5
.
9
3
3
.
0
0
5
.
7
1
2
2
.
6
6
2
2
.
6
8
6
.
0
9
1
8
.
1
0
1
8
.
1
6
6
.
3
4
1
5
.
1
9
1
5
.
2
6
5
.
0
0
6
.
3
6
2
0
.
6
0
2
0
.
6
1
6
.
5
3
1
7
.
0
1
1
7
.
0
6
6
.
6
4
1
4
.
5
8
1
4
.
6
4
7
.
0
0
7
.
0
1
1
8
.
9
0
1
8
.
9
1
6
.
9
8
1
6
.
0
5
1
6
.
1
0
6
.
9
5
1
4
.
0
2
1
4
.
0
8
9
.
0
0
7
.
6
7
1
7
.
4
7
1
7
.
4
8
7
.
4
3
1
5
.
2
0
1
5
.
2
5
7
.
2
6
1
3
.
5
0
1
3
.
5
6
1
0
.
0
0
8
.
0
0
1
6
.
8
3
1
6
.
8
4
7
.
6
5
1
4
.
8
1
1
4
.
8
6
7
.
4
1
1
3
.
2
6
1
3
.
3
1
50 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
T
a
b
l
e
A
.
1
c
o
n
t
i
n
u
e
d
M
a
t
u
r
i
t
y
o
f
t
h
e
v
a
r
i
a
n
c
e
c
o
n
t
r
a
c
t
T
e
n
o
r
=
1
0
y
e
a
r
s
1
m
o
n
t
h
6
m
o
n
t
h
s
1
2
m
o
n
t
h
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
1
.
0
0
5
.
2
5
1
7
.
8
0
1
7
.
8
0
5
.
4
8
1
5
.
9
2
1
5
.
9
3
5
.
7
2
1
4
.
1
2
1
4
.
1
4
3
.
0
0
5
.
8
1
1
6
.
5
1
1
6
.
5
1
5
.
9
7
1
4
.
9
6
1
4
.
9
7
6
.
1
3
1
3
.
4
3
1
3
.
4
5
5
.
0
0
6
.
3
9
1
5
.
4
2
1
5
.
4
1
6
.
4
7
1
4
.
1
3
1
4
.
1
4
6
.
5
4
1
2
.
8
2
1
2
.
8
4
7
.
0
0
6
.
9
8
1
4
.
4
8
1
4
.
4
8
6
.
9
7
1
3
.
4
0
1
3
.
4
1
6
.
9
6
1
2
.
2
7
1
2
.
2
9
9
.
0
0
7
.
5
9
1
3
.
6
7
1
3
.
6
7
7
.
4
9
1
2
.
7
5
1
2
.
7
7
7
.
3
9
1
1
.
7
8
1
1
.
8
0
1
0
.
0
0
7
.
9
1
1
3
.
3
1
1
3
.
3
1
7
.
7
6
1
2
.
4
6
1
2
.
4
7
7
.
6
1
1
1
.
5
5
1
1
.
5
7
M
a
t
u
r
i
t
y
o
f
t
h
e
v
a
r
i
a
n
c
e
c
o
n
t
r
a
c
t
T
e
n
o
r
=
2
0
y
e
a
r
s
1
m
o
n
t
h
6
m
o
n
t
h
s
1
2
m
o
n
t
h
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
1
.
0
0
5
.
7
9
1
0
.
5
5
1
0
.
5
5
5
.
9
5
6
1
9
.
6
5
9
.
6
6
6
.
1
1
8
.
7
4
8
.
7
6
3
.
0
0
6
.
1
7
1
0
.
2
7
1
0
.
2
7
6
.
2
7
6
5
9
.
4
4
9
.
4
5
6
.
3
8
8
.
5
9
8
.
6
0
5
.
0
0
6
.
5
5
1
0
.
0
3
1
0
.
0
3
6
.
6
0
6
7
9
.
2
4
9
.
2
6
6
.
6
5
8
.
4
4
8
.
4
6
7
.
0
0
6
.
9
5
9
.
7
9
9
.
7
9
6
.
9
4
7
0
9
.
0
6
9
.
0
7
6
.
9
3
8
.
3
0
8
.
3
2
9
.
0
0
7
.
3
6
9
.
5
7
9
.
5
7
7
.
2
9
7
8
8
.
8
8
8
.
9
0
7
.
2
2
8
.
1
7
8
.
1
8
1
0
.
0
0
7
.
5
7
9
.
4
7
9
.
4
7
7
.
4
7
7
1
8
.
8
1
8
.
8
2
7
.
3
7
8
.
1
0
8
.
1
2
M
a
t
u
r
i
t
y
o
f
t
h
e
v
a
r
i
a
n
c
e
c
o
n
t
r
a
c
t
T
e
n
o
r
=
3
0
y
e
a
r
s
1
m
o
n
t
h
6
m
o
n
t
h
s
1
2
m
o
n
t
h
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
E
-
V
o
l
1
.
0
0
5
.
9
6
8
.
7
0
8
.
7
0
6
.
0
9
2
8
8
.
0
1
8
.
0
2
6
.
2
2
7
.
3
0
7
.
3
1
3
.
0
0
6
.
2
7
8
.
5
9
8
.
5
9
6
.
3
6
5
1
7
.
9
3
7
.
9
3
6
.
4
5
7
.
2
3
7
.
2
5
5
.
0
0
6
.
6
0
8
.
4
9
8
.
4
9
6
.
6
4
6
6
7
.
8
4
7
.
8
5
6
.
6
9
7
.
1
7
7
.
1
9
7
.
0
0
6
.
9
4
8
.
3
9
8
.
3
9
6
.
9
3
7
4
7
.
7
6
7
.
7
7
6
.
9
3
7
.
1
1
7
.
1
3
9
.
0
0
7
.
2
9
8
.
2
9
8
.
2
9
7
.
2
3
7
8
7
.
6
8
7
.
6
9
7
.
1
7
7
.
0
5
7
.
0
7
1
0
.
0
0
7
.
4
7
8
.
2
4
8
.
2
4
7
.
3
9
1
7
7
.
6
5
7
.
6
6
7
.
3
0
7
.
0
2
7
.
0
4
51 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
The left panels of Figure 7 depict the approximations to the square root of the two expectations
in Eq. (F.5): the rst, normalized by IVLI
t
(r
t
; T
1
, , T
a
) (T t), as in Eq. (25); and the second,
normalized by (T t), as in Eq. (26). The right panels of Figure 7 depict the instantaneous volatility
of the forward swap rate, as dened in Eq. (F.4). Finally, Eqs. (F.3), (F.4) and (F.5) are evaluated
assuming reset dates are quarterly and, accordingly, setting c
i
constant and equal to
1
4
.
G. Spanning and hedging Gaussian contracts
G.1. Pricing
We price contracts where the variance payo in Eq. (A.5) is replaced by:
\
BP
a
(t, T) IVLI
T
(T
1
, , T
a
) ,
where \
BP
a
(t, T) denotes the realized variance of the forward swap changes in the interval [t, T[,
\
BP
a
(t, T) =
_
T
t
1
2
c
(T
1
, , T
a
) o
2
c
(T
1
, , T
a
) d:, (G.1)
and o
2
c
(T
1
, , T
a
) is as in Eq. (A.6). By a Taylors expansion with remainder, we have, denoting, as
usual, for simplicity, 1
t
= 1
t
(T
1
, , T
a
),
1
2
T
= 1
2
t
21
t
(1
T
1
t
) 2
__
1t
0
(1 1
T
)
+
d1
_
o
1t
(1
T
1)
+
d1
_
. (G.2)
Multiplying both sides of this equation by IVLI
t
(T
1
, , T
a
), and taking expectations under the
Q
swap
probability, leaves:
IVLI
t
(T
1
, , T
a
) E
Qswap
_
1
2
T
1
2
t
_
= 2
__
1t
0
SWIN
R
t
(1, T; T
1
, , T
a
) d1
_
o
1t
SWIN
P
t
(1, T; T
1
, , T
a
) d1
_
. (G.3)
Moreover, by Its lemma, and Eqs. (A.4) and (G.1),
E
Qswap
_
1
2
T
1
2
t
_
= E
Qswap
_
\
BP
a
(t, T)

.
By replacing this expression into Eq. (G.3) yields:
2
__
1t
0
SWIN
R
t
(1, T; T
1
, , T
a
) d1
_
o
1t
SWIN
P
t
(1, T; T
1
, , T
a
) d1
_
= IVLI
t
(T
1
, , T
a
) E
Qswap
_
1
2
T
1
2
t
_
= IVLI
t
(T
1
, , T
a
) E
Qswap
_
\
BP
a
(t, T)

= E
Q
_
c

_
T
t
vsoc
_
IVLI
T
(T
1
, , T
a
) \
BP
a
(t, T)
_
_
= F
BP
var,a
(t, T) , (G.4)
where the last line follows by the Radon-Nikodym derivative dened in Eq. (A.3).
The basis point volatility in Eq. (27) can also be simplied using the Blacks formulae in Eqs. (B.1)
and (B.10) to compute F
BP
var,a
(t, T) in Eq. (G.4), as follows:
IIS-VI
BP
a
(t, T)
52 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
=

_
2
T t
_
_

i:1
i
<1t

7
t
_
1
a
t
, T, 1
i
; (T t) IV
2
i,t
_
^1
i


i:1
i
1t
7
t
_
1
a
t
, T, 1
i
; (T t) IV
2
i,t
_
^1
i
_
_
,
(G.5)
where 7
t
and

7
t
are as in Eqs. (B.1) and (B.10), 1
a
t
is the current forward swap rate for maturity T
and tenor length T
a
T, 1
a
t
= 1
t
(T
1
, , T
a
), and, nally, IV
i,t
denotes the time t implied percentage
volatility for swaptions with strike equal to 1
i
.
G.2. Hedging
We provide the details leading to Table III only, as those for Table IV are nearly identical. By Its
lemma:
IVLI
T
(T
1
, , T
a
) \
BP
a
(t, T)
= 2IVLI
T
(T
1
, , T
a
)
_
T
t
1
c
(T
1
, , T
a
) d1
c
(T
1
, , T
a
)
IVLI
T
(T
1
, , T
a
)
_
1
2
T
(T
1
, , T
a
) 1
2
t
(T
1
, , T
a
)
_
. (G.6)
By Eq. (G.2), the second term on the right hand side is the payo at T of a portfolio set up at t, which
is: (a) long 21
t
(T
1
, , T
a
) units a xed interest payer swap struck at 1
t
(T
1
, , T
a
), and (b) long a
continuum of out-of-the-money swaptions with weights 2d1. It is the static position (ii) in Table III
of the main text. By Eqs. (G.2), (G.3) and (G.4), and the denition of the Radon-Nikodym derivative
in Eq. (A.3), its cost is F
BP
var,a
(t, T), which we borrow at t, to repay it back at T, as in row (iii) of
Table III. The self-nanced portfolio to be shorted, as indicated by row (i) of Table III, is obtained
similarly as the portfolio in row (i) of Table I (see Appendix C.3), but with the portfolio,

c
t
'
t
= IVLI
t
(T
1
, , T
a
)
__
t
t
21
c
(T
1
, , T
a
) d1
c
(T
1
, , T
a
) 1
_
,
replacing that in Eq. (C.8).
G.3. Constant Maturity Swaps
Consider, initially, the fair price of the payment of a Constant Maturity Swap (CMS) occurring at
time T
0
i, and set, for simplicity, o (T
0
) = 1
T
0
(T
1
, , T
a
). The current value of o (T
0
) to be paid
at time T
0
i is the same as the current value of o (T
0
) 1
T
0
(T
0
i) to be paid at T
0
, such that:
cms (t, T
0
i) = E
Q
_
c

_
T
0
t
vuo&
o (T
0
) 1
T
0
(T
0
i)
_
= 1
t
(T
0
i) E
Qswap
_
o (T
0
)
(
T
0
(
t
_
, (G.7)
where (
t
=
1 (T
0
+i)
PVBP (T
1
, ,Tn)
. We calculate (
t
, by discounting through the reset times, T
i
T
0
, using
the at rate formula, 1
T
0
(T
i
) = (1 co (T
0
))
i
, where c is the year fraction between the reset times,
such that,
(
t
=
1
t
(T
0
i)
IVLI
t
(T
1
, , T
a
)
-
(1 co (t))
(T
0
+it)
c

a
i=1
(1 co (t))
(T
i
t)
=
o (t) (1 co (t))
i
1
1
(1+cS(t))
n
= G(o (t)) . (G.8)
The previous derivations closely follow those in Hagan (2003), and are provided for completeness. We
now make the connection between the price of the CMS in Eq. (G.7) and the price of the IRV forward
contract in BP of Denition IV-(a) in the main text. Replacing the approximation in Eq. (G.8) into
Eq. (G.7) leaves:
cms (t, T
0
i)
53 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
= 1
t
(T
0
i) E
swap
_
o (T
0
)
G(o (T
0
))
G(1
t
)
_
- 1
t
(T
0
i) E
swap
_
o (T
0
)
_
1
G
t
(1
t
)
G(1
t
)
(o (T
0
) 1
t
)
__
= 1
t
(T
0
i) E
swap
(o (T
0
)) IVLI
t
(T
1
, , T
a
) G
t
(1
t
) E
Qswap
[o (T
0
) (o (T
0
) 1
t
)[
= 1
t
(T
0
i) 1
t
G
t
(1
t
) IVLI
t
(T
1
, , T
a
) E
Qswap
_
o
2
(T
0
) 1
2
t
_
= 1
t
(T
0
i) 1
t
G
t
(1
t
) F
BP
var,a
(t, T
0
) ,
where the second line follows by a rst order Taylor approximation of the function G about 1
t
, the
third from the denition of G(1
t
), the fourth from the martingale property of the forward swap rate
under Q
swap
, E
swap
(o (T
0
)) = 1
t
and, nally, the fth equality from Eq. (G.3) and Eq. (G.4). Eq. (18)
of the main text follows by calculating and summing every single CMS payment, cms (t, T
0
,i), for
, = 1, , .
As mentioned in Section 5.3 of the main text, it is well-known since at least Hagan (2003) and
Mercurio and Pallavicini (2006) that CMS link to the entire skew. However, our representation of
the price of a CMS in Eq. (18) quite diers from previous ones. Mercurio and Pallavicini (2006), for
example, utilize spanning arguments dierent from ours. We explain the dierences.
Consider a Taylors expansion with remainder of the function ) (1
T
) = 1
2
T
about some point 1
c
,
1
2
T
= 1
2
c
21
c
(1
T
1
c
) 2
__
1o
0
(1 1
T
)
+
d1
_
o
1o
(1
T
1)
+
d1
_
. (G.9)
Mercurio and Pallavicini (2006) consider the point 1
c
= 0, such that, Eq. (G.9) collapses to
1
2
T
= 2
_
o
0
(1
T
1)
+
d1,
such that
E
Qswap
_
1
2
T
_
=
2
IVLI
t
(T
1
, , T
a
)
_
o
0
SWIN
P
t
(1, T; T
a
) d1.
Our approach diers as we take 1
c
= 1
t
in Eq. (G.9), leading to Eq. (G.2) and, then, to,
E
Qswap
_
1
2
T
1
2
t
_
= 2E
Qswap
__
1t
0
(1 1
T
)
+
d1
_
o
1t
(1
T
1)
+
d1
_
=
2
IVLI
t
(T
1
, , T
a
)
__
1t
0
SWIN
R
t
(1, T; T
a
) d1
_
o
1t
SWIN
P
t
(1, T; T
a
) d1
_
.
G.4. Numerical experiments
Table A.2 reports experiments relating to the calculation of BP volatility as referenced by the index
IIS-VI
BP
a
(t, T) in Eq. (33), as well as future expected volatility in a risk-neutral market, as predicted
by the Vasicek model, using the same parameter values as those in Appendix F.
54 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
T
a
b
l
e
A
.
2
T
h
e
I
I
S
-
V
I
B
P
a
(
t
,
T
)
i
n
d
e
x
i
n
E
q
.
(
3
3
)
(
l
a
b
e
l
e
d
I
R
S
-
V
I
B
P
)
,
a
n
d
f
u
t
u
r
e
e
x
p
e
c
t
e
d
v
o
l
a
t
i
l
i
t
y
i
n
a
r
i
s
k
-
n
e
u
t
r
a
l
m
a
r
k
e
t
,
_
1
T

t
1
Q
[
\
B
P
a
(
t
,
T
)
[
,
w
h
e
r
e
\
B
P
a
(
t
,
T
)
i
s
a
s
i
n
E
q
.
(
G
.
1
)
(
l
a
b
e
l
e
d
E
-
V
o
l
B
P
)
,
f
o
r
(
i
)
t
i
m
e
-
t
o
m
a
t
u
r
i
t
y
T

t
e
q
u
a
l
t
o
1
,
6
a
n
d
9
m
o
n
t
h
s
,
a
n
d
1
,
2
a
n
d
3
y
e
a
r
s
,
a
n
d
(
i
i
)
t
e
n
o
r
l
e
n
g
t
h
:
e
q
u
a
l
t
o
1
,
2
,
3
a
n
d
5
y
e
a
r
s
,
a
n
d
f
o
r
t
i
m
e
-
t
o
m
a
t
u
r
i
t
y
T

t
e
q
u
a
l
t
o
1
,
6
a
n
d
1
2
m
o
n
t
h
s
,
a
n
d
(
i
i
)
t
e
n
o
r
l
e
n
g
t
h
:
e
q
u
a
l
t
o
1
0
,
2
0
a
n
d
3
0
y
e
a
r
s
.
T
h
e
i
n
d
e
x
a
n
d
e
x
p
e
c
t
e
d
v
o
l
a
t
i
l
i
t
i
e
s
a
r
e
c
o
m
p
u
t
e
d
u
n
d
e
r
t
h
e
a
s
s
u
m
p
t
i
o
n
t
h
e
s
h
o
r
t
-
t
e
r
m
i
n
t
e
r
e
s
t
r
a
t
e
i
s
a
s
i
n
t
h
e
V
a
s
i
c
e
k
m
o
d
e
l
o
f
E
q
.
(
F
.
1
)
,
w
i
t
h
p
a
r
a
m
e
t
e
r
v
a
l
u
e
s
c
=
0
.
8
8
0
7
,
r
=
0
.
0
7
2
a
n
d
o

=
0
.
0
8
8
1
.
M
a
t
u
r
i
t
y
o
f
t
h
e
v
a
r
i
a
n
c
e
c
o
n
t
r
a
c
t
T
e
n
o
r
=
1
y
e
a
r
1
m
o
n
t
h
6
m
o
n
t
h
s
9
m
o
n
t
h
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
1
.
0
0
2
.
1
8
2
7
2
.
4
3
2
7
2
.
4
3
2
.
9
0
2
5
2
.
9
2
2
5
2
.
9
7
3
.
2
7
2
4
2
.
3
3
2
4
2
.
4
1
3
.
0
0
3
.
8
0
2
7
3
.
7
9
2
7
3
.
7
9
4
.
2
8
2
5
3
.
9
9
2
5
4
.
0
5
4
.
5
3
2
4
3
.
2
6
2
4
3
.
3
6
5
.
0
0
5
.
4
4
2
7
5
.
1
5
2
7
5
.
1
6
5
.
6
8
2
5
5
.
0
7
2
5
5
.
1
3
5
.
8
0
2
4
4
.
1
9
2
4
4
.
3
0
7
.
0
0
7
.
0
8
2
7
6
.
5
2
2
7
6
.
5
4
7
.
0
8
2
5
6
.
1
5
2
5
6
.
2
2
7
.
0
8
2
4
5
.
1
3
2
4
5
.
2
6
9
.
0
0
8
.
7
3
2
7
7
.
9
0
2
7
7
.
9
2
8
.
4
9
2
5
7
.
2
3
2
5
7
.
3
2
8
.
3
6
2
4
6
.
0
7
2
4
6
.
2
1
1
0
.
0
0
9
.
5
6
2
7
8
.
5
9
2
7
8
.
6
2
9
.
1
9
2
5
7
.
7
8
2
5
7
.
8
7
9
.
0
0
2
4
6
.
5
4
2
4
6
.
6
9
T
e
n
o
r
=
2
y
e
a
r
s
1
m
o
n
t
h
6
m
o
n
t
h
s
9
m
o
n
t
h
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
1
.
0
0
2
.
9
3
2
2
9
.
6
0
2
2
9
.
6
0
3
.
5
3
2
1
3
.
3
0
2
1
3
.
3
7
3
.
8
4
2
0
4
.
4
5
2
0
4
.
5
5
3
.
0
0
4
.
3
1
2
3
1
.
1
5
2
3
1
.
1
5
4
.
7
0
2
1
4
.
5
3
2
1
4
.
6
0
4
.
9
1
2
0
5
.
5
1
2
0
5
.
6
3
5
.
0
0
5
.
6
8
2
3
2
.
7
1
2
3
2
.
7
2
5
.
8
8
2
1
5
.
7
6
2
1
5
.
8
4
5
.
9
8
2
0
6
.
5
7
2
0
6
.
7
2
7
.
0
0
7
.
0
7
2
3
4
.
2
9
2
3
4
.
3
0
7
.
0
7
2
1
7
.
0
0
2
1
7
.
0
9
7
.
0
6
2
0
7
.
6
5
2
0
7
.
8
1
9
.
0
0
8
.
4
7
2
3
5
.
8
8
2
3
5
.
8
9
8
.
2
6
2
1
8
.
2
5
2
1
8
.
3
5
8
.
1
5
2
0
8
.
7
3
2
0
8
.
9
0
1
0
.
0
0
9
.
1
8
2
3
6
.
6
8
2
3
6
.
6
9
8
.
8
6
2
1
8
.
8
7
2
1
8
.
9
9
8
.
6
9
2
0
9
.
2
8
2
0
9
.
4
6
T
e
n
o
r
=
3
y
e
a
r
s
1
m
o
n
t
h
6
m
o
n
t
h
s
9
m
o
n
t
h
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
1
.
0
0
3
.
5
1
1
9
6
.
1
3
1
9
6
.
1
4
4
.
0
2
1
8
2
.
4
1
1
8
2
.
4
8
4
.
2
8
1
7
4
.
9
3
1
7
5
.
0
5
3
.
0
0
4
.
6
8
1
9
7
.
9
7
1
9
7
.
9
7
5
.
0
2
1
8
3
.
8
5
1
8
3
.
9
3
5
.
2
0
1
7
6
.
1
8
1
7
6
.
3
2
5
.
0
0
5
.
8
7
1
9
9
.
8
2
1
9
9
.
8
2
6
.
0
3
1
8
5
.
3
1
1
8
5
.
4
0
6
.
1
2
1
7
7
.
4
5
1
7
7
.
6
0
7
.
0
0
7
.
0
6
2
0
1
.
6
8
2
0
1
.
6
9
7
.
0
5
1
8
6
.
7
8
1
8
6
.
8
8
7
.
0
5
1
7
8
.
7
2
1
7
8
.
9
0
9
.
0
0
8
.
2
7
2
0
3
.
5
7
2
0
3
.
5
8
8
.
0
8
1
8
8
.
2
6
1
8
8
.
3
8
7
.
9
8
1
8
0
.
0
1
1
8
0
.
2
0
1
0
.
0
0
8
.
8
7
2
0
4
.
5
2
2
0
4
.
5
3
8
.
6
0
1
8
9
.
0
1
1
8
9
.
1
3
8
.
4
5
1
8
0
.
6
6
1
8
0
.
8
5
T
e
n
o
r
=
5
y
e
a
r
s
1
m
o
n
t
h
6
m
o
n
t
h
s
9
m
o
n
t
h
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
1
.
0
0
4
.
3
1
1
4
9
.
3
1
1
4
9
.
3
1
4
.
6
9
1
3
9
.
1
9
1
3
9
.
2
7
4
.
8
9
1
3
3
.
6
6
1
3
3
.
7
9
3
.
0
0
5
.
2
0
1
5
1
.
5
7
1
5
1
.
5
8
5
.
4
6
1
4
0
.
9
8
1
4
1
.
0
7
5
.
5
9
1
3
5
.
2
2
1
3
5
.
3
6
5
.
0
0
6
.
1
1
1
5
3
.
8
7
1
5
3
.
8
7
6
.
2
3
1
4
2
.
7
9
1
4
2
.
9
0
6
.
3
0
1
3
6
.
8
0
1
3
6
.
9
6
7
.
0
0
7
.
0
3
1
5
6
.
2
0
1
5
6
.
2
1
7
.
0
2
1
4
4
.
6
3
1
4
4
.
7
5
7
.
0
1
1
3
8
.
3
9
1
3
8
.
5
7
9
.
0
0
7
.
9
7
1
5
8
.
5
6
1
5
8
.
5
7
7
.
8
2
1
4
6
.
4
9
1
4
6
.
6
1
7
.
7
4
1
4
0
.
0
1
1
4
0
.
2
0
1
0
.
0
0
8
.
4
4
1
5
9
.
7
6
1
5
9
.
7
6
8
.
2
2
1
4
7
.
4
3
1
4
7
.
5
6
8
.
1
1
1
4
0
.
8
2
1
4
1
.
0
2
55 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
T
a
b
l
e
A
.
2
c
o
n
t
i
n
u
e
d
M
a
t
u
r
i
t
y
o
f
t
h
e
v
a
r
i
a
n
c
e
c
o
n
t
r
a
c
t
T
e
n
o
r
=
1
y
e
a
r
1
y
e
a
r
2
y
e
a
r
s
3
y
e
a
r
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
1
.
0
0
3
.
6
1
2
3
2
.
5
2
2
3
2
.
6
2
4
.
6
7
1
9
9
.
6
5
1
9
9
.
9
1
5
.
3
8
1
7
4
.
9
8
1
7
5
.
3
3
3
.
0
0
4
.
7
6
2
3
3
.
3
2
2
3
3
.
4
4
5
.
4
5
2
0
0
.
1
2
2
0
0
.
3
9
5
.
9
2
1
7
5
.
2
5
1
7
5
.
6
2
5
.
0
0
5
.
9
1
2
3
4
.
1
3
2
3
4
.
2
7
6
.
2
4
2
0
0
.
5
8
2
0
0
.
8
8
6
.
4
6
1
7
5
.
5
3
1
7
5
.
9
1
7
.
0
0
7
.
0
7
2
3
4
.
9
5
2
3
5
.
1
0
7
.
0
4
2
0
1
.
0
5
2
0
1
.
3
6
7
.
0
0
1
7
5
.
8
1
1
7
6
.
2
1
9
.
0
0
8
.
2
3
2
3
5
.
7
7
2
3
5
.
9
3
7
.
8
3
2
0
1
.
5
3
2
0
1
.
8
5
7
.
5
4
1
7
6
.
0
9
1
7
6
.
5
0
1
0
.
0
0
8
.
8
2
2
3
6
.
1
8
2
3
6
.
3
5
8
.
2
3
2
0
1
.
7
6
2
0
2
.
1
0
7
.
8
1
1
7
6
.
2
3
1
7
6
.
6
4
M
a
t
u
r
i
t
y
o
f
t
h
e
v
a
r
i
a
n
c
e
c
o
n
t
r
a
c
t
T
e
n
o
r
=
2
y
e
a
r
s
1
y
e
a
r
2
y
e
a
r
s
3
y
e
a
r
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
1
.
0
0
4
.
1
3
1
9
6
.
2
3
1
9
6
.
3
6
5
.
0
1
1
6
8
.
6
7
1
6
8
.
9
5
5
.
6
1
1
4
7
.
9
5
1
4
8
.
3
0
3
.
0
0
5
.
1
0
1
9
7
.
1
6
1
9
7
.
3
0
5
.
6
8
1
6
9
.
2
1
1
6
9
.
5
1
6
.
0
7
1
4
8
.
2
6
1
4
8
.
6
4
5
.
0
0
6
.
0
7
1
9
8
.
0
9
1
9
8
.
2
5
6
.
3
5
1
6
9
.
7
4
1
7
0
.
0
7
6
.
5
2
1
4
8
.
5
8
1
4
8
.
9
7
7
.
0
0
7
.
0
5
1
9
9
.
0
3
1
9
9
.
2
0
7
.
0
2
1
7
0
.
2
9
1
7
0
.
6
2
6
.
9
8
1
4
8
.
9
1
1
4
9
.
3
0
9
.
0
0
8
.
0
4
1
9
9
.
9
8
2
0
0
.
1
6
7
.
6
9
1
7
0
.
8
3
1
7
1
.
1
9
7
.
4
4
1
4
9
.
2
3
1
4
9
.
6
4
1
0
.
0
0
8
.
5
3
2
0
0
.
4
5
2
0
0
.
6
4
8
.
0
3
1
7
1
.
1
1
1
7
1
.
4
7
7
.
6
7
1
4
9
.
3
9
1
4
9
.
8
1
M
a
t
u
r
i
t
y
o
f
t
h
e
v
a
r
i
a
n
c
e
c
o
n
t
r
a
c
t
T
e
n
o
r
=
3
y
e
a
r
s
1
y
e
a
r
2
y
e
a
r
s
3
y
e
a
r
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
1
.
0
0
4
.
5
2
1
6
7
.
9
9
1
6
8
.
1
3
5
.
2
8
1
4
4
.
6
2
1
4
4
.
9
1
5
.
7
9
1
2
7
.
0
0
1
2
7
.
3
5
3
.
0
0
5
.
3
6
1
6
9
.
0
9
1
6
9
.
2
4
5
.
8
5
1
4
5
.
2
6
1
4
5
.
5
7
6
.
1
8
1
2
7
.
3
8
1
2
7
.
7
5
5
.
0
0
6
.
1
9
1
7
0
.
1
9
1
7
0
.
3
6
6
.
4
2
1
4
5
.
9
0
1
4
6
.
2
3
6
.
5
7
1
2
7
.
7
6
1
2
8
.
1
4
7
.
0
0
7
.
0
4
1
7
1
.
3
1
1
7
1
.
4
9
7
.
0
1
1
4
6
.
5
5
1
4
6
.
8
9
6
.
9
7
1
2
8
.
1
4
1
2
8
.
5
4
9
.
0
0
7
.
8
9
1
7
2
.
4
3
1
7
2
.
6
3
7
.
5
8
1
4
7
.
2
0
1
4
7
.
5
6
7
.
3
6
1
2
8
.
5
3
1
2
8
.
9
3
1
0
.
0
0
8
.
3
1
1
7
2
.
9
9
1
7
3
.
2
0
7
.
8
7
1
4
7
.
5
3
1
4
7
.
8
9
7
.
5
6
1
2
8
.
7
3
1
2
9
.
1
3
M
a
t
u
r
i
t
y
o
f
t
h
e
v
a
r
i
a
n
c
e
c
o
n
t
r
a
c
t
T
e
n
o
r
=
5
y
e
a
r
s
1
y
e
a
r
2
y
e
a
r
s
3
y
e
a
r
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
1
.
0
0
5
.
0
7
1
2
8
.
5
1
1
2
8
.
6
6
5
.
6
5
1
1
1
.
0
4
1
1
1
.
3
3
6
.
0
4
9
7
.
7
5
9
8
.
0
9
3
.
0
0
5
.
7
1
1
2
9
.
8
8
1
3
0
.
0
4
6
.
0
9
1
1
1
.
8
4
1
1
2
.
1
4
6
.
3
4
9
8
.
2
3
9
8
.
5
8
5
.
0
0
6
.
3
6
1
3
1
.
2
5
1
3
1
.
4
3
6
.
5
3
1
1
2
.
6
4
1
1
2
.
9
6
6
.
6
4
9
8
.
7
1
9
9
.
0
7
7
.
0
0
7
.
0
1
1
3
2
.
6
5
1
3
2
.
8
4
6
.
9
8
1
1
3
.
4
5
1
1
3
.
7
9
6
.
9
5
9
9
.
2
0
9
9
.
5
6
9
.
0
0
7
.
6
7
1
3
4
.
0
6
1
3
4
.
2
6
7
.
4
3
1
1
4
.
2
7
1
1
4
.
6
2
7
.
2
6
9
9
.
6
9
1
0
0
.
0
6
1
0
.
0
0
8
.
0
0
1
3
4
.
7
7
1
3
4
.
9
7
7
.
6
5
1
1
4
.
6
8
1
1
5
.
0
4
7
.
4
1
9
9
.
9
3
1
0
0
.
3
1
56 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
T
a
b
l
e
A
.
2
c
o
n
t
i
n
u
e
d
M
a
t
u
r
i
t
y
o
f
t
h
e
v
a
r
i
a
n
c
e
c
o
n
t
r
a
c
t
T
e
n
o
r
=
1
0
y
e
a
r
s
1
m
o
n
t
h
6
m
o
n
t
h
s
1
2
m
o
n
t
h
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
1
.
0
0
5
.
2
5
9
3
.
4
9
9
3
.
5
0
5
.
4
8
8
7
.
5
8
8
7
.
6
5
5
.
7
2
9
9
8
1
.
2
6
8
1
.
3
9
3
.
0
0
5
.
8
1
9
6
.
0
0
9
6
.
0
0
5
.
9
7
8
9
.
5
7
8
9
.
6
5
6
.
1
3
5
0
8
2
.
7
8
8
2
.
9
3
5
.
0
0
6
.
3
9
9
8
.
5
6
9
8
.
5
6
6
.
4
7
9
1
.
6
0
9
1
.
6
9
6
.
5
4
7
7
8
4
.
3
3
8
4
.
4
9
7
.
0
0
6
.
9
8
1
0
1
.
1
8
1
0
1
.
1
9
6
.
9
7
9
3
.
6
7
9
3
.
7
7
6
.
9
6
8
2
8
5
.
9
1
8
6
.
0
7
9
.
0
0
7
.
5
9
1
0
3
.
8
6
1
0
3
.
8
7
7
.
4
9
9
5
.
7
8
9
5
.
8
9
7
.
3
9
6
5
8
7
.
5
1
8
7
.
6
8
1
0
.
0
0
7
.
9
0
1
0
5
.
2
3
1
0
5
.
2
3
7
.
7
6
9
6
.
8
5
9
6
.
9
6
7
.
6
1
3
7
8
8
.
3
2
8
8
.
4
9
M
a
t
u
r
i
t
y
o
f
t
h
e
v
a
r
i
a
n
c
e
c
o
n
t
r
a
c
t
T
e
n
o
r
=
2
0
y
e
a
r
s
1
m
o
n
t
h
6
m
o
n
t
h
s
1
2
m
o
n
t
h
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
1
.
0
0
5
.
7
9
9
8
6
1
.
2
1
6
1
.
2
1
5
.
9
5
5
7
.
5
6
5
7
.
6
2
6
.
1
1
4
6
5
3
.
6
2
5
3
.
7
2
3
.
0
0
6
.
1
7
0
5
6
3
.
4
3
6
3
.
4
3
6
.
2
7
5
9
.
3
3
5
9
.
4
0
6
.
3
8
2
7
5
4
.
9
9
5
5
.
1
0
5
.
0
0
6
.
5
5
4
6
6
5
.
7
2
6
5
.
7
2
6
.
6
0
6
1
.
1
5
6
1
.
2
3
6
.
6
5
7
7
5
6
.
3
8
5
6
.
5
0
7
.
0
0
6
.
9
5
2
6
6
8
.
0
9
6
8
.
0
9
6
.
9
4
6
3
.
0
2
6
3
.
1
0
6
.
9
3
9
6
5
7
.
8
0
5
7
.
9
3
9
.
0
0
7
.
3
6
4
9
7
0
.
5
3
7
0
.
5
3
7
.
2
9
6
4
.
9
4
6
5
.
0
3
7
.
2
2
8
7
5
9
.
2
5
5
9
.
3
9
1
0
.
0
0
7
.
5
7
6
6
7
1
.
7
7
7
1
.
7
8
7
.
4
7
6
5
.
9
2
6
6
.
0
1
7
.
3
7
5
9
5
9
.
9
9
6
0
.
1
3
M
a
t
u
r
i
t
y
o
f
t
h
e
v
a
r
i
a
n
c
e
c
o
n
t
r
a
c
t
T
e
n
o
r
=
3
0
y
e
a
r
s
1
m
o
n
t
h
6
m
o
n
t
h
s
1
2
m
o
n
t
h
s
S
h
o
r
t
T
e
r
m
R
a
t
e
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
F
o
r
w
a
r
d
S
w
a
p
R
a
t
e
I
R
S
-
V
I
B
P
E
-
V
o
l
B
P
1
.
0
0
5
.
9
6
5
1
.
9
1
5
1
.
9
1
6
.
0
9
4
8
.
8
8
4
8
.
9
3
6
.
2
2
4
5
.
5
9
4
5
.
6
8
3
.
0
0
6
.
2
7
5
3
.
9
5
5
3
.
9
5
6
.
3
6
5
0
.
5
1
5
0
.
5
7
6
.
4
5
4
6
.
8
5
4
6
.
9
4
5
.
0
0
6
.
6
0
5
6
.
0
6
5
6
.
0
7
6
.
6
4
5
2
.
1
9
5
2
.
2
5
6
.
6
9
4
8
.
1
3
4
8
.
2
4
7
.
0
0
6
.
9
4
5
8
.
2
5
5
8
.
2
5
6
.
9
3
5
3
.
9
1
5
3
.
9
8
6
.
9
3
4
9
.
4
5
4
9
.
5
6
9
.
0
0
7
.
2
9
6
0
.
5
1
6
0
.
5
1
7
.
2
3
5
5
.
6
9
5
5
.
7
7
7
.
1
7
5
0
.
7
9
5
0
.
9
1
1
0
.
0
0
7
.
4
7
6
1
.
6
6
6
1
.
6
6
7
.
3
9
5
6
.
5
9
5
6
.
6
8
7
.
3
0
5
1
.
4
7
5
1
.
6
0
57 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
G.5. Constant vega in Gaussian markets
We show that the statement in (17) is trueconstant vega requires swaptions to be equally weighted
in a Gaussian market. If the forward swap rate is solution to Eq. (13), the price of a swaption payer
is:
O
P
t
(1
t
(T
1
, , T
a
) , 1, T, o
.
; T
a
) = IVLI
t
(T
1
, , T
a
) ?
P
t
(1
t
(T
1
, , T
a
) , 1, T, o
.
; T
a
) ,
where
?
P
t
(1, 1, T, o; T
a
) = (1 1) 1
_
1 1
o
_
T t
_
o
_
T tc
_
1 1
o
_
T t
_
,
and c denotes the standard normal density. The vega of a payer is the same as the vega of a receiver,
and equals:
i
C
t
(1, o) =
0O
P
t
(1, 1, T, o; T
a
)
0o
= IVLI
t
(T
1
, , T
a
)
_
T tc
_
1 1
o
_
T t
_
,
such that the vega of a portfolio of swaptions (be they payers and/or receivers) is:
i
t
(1, o) =
0
t
(1, o)
0o
= IVLI
t
(T
1
, , T
a
)
_
T t
_
. (1) c
_
1 1
o
_
T t
_
d1. (G.10)
As for the if part in (17), let . (1) = consl., such that by Eq. (G.10), and the obvious fact that the
Gaussian density c integrates to one, we have indeed that the vega is independent of 1,
i
t
(1, o) = IVLI
t
(T
1
, , T
a
)
_
T t consl..
As for the only if part, let us dierentiate i
t
(1, o) in Eq. (G.10) with respect to 1,
0i
t
(1, o)
01
=
IVLI
t
(T
1
, , T
a
)
o
2
_
T t
_
. (1) c
_
1 1
o
_
T t
_
(1 1) d1.
The only function . which is independent of 1, and such that
0it(1,o)
01
is identically zero, is the constant
weighting.
H. Jumps
We derive the fair value of the Standardized IRV swap rate in Denition III, and the Standardized
BP-IRV swap rate of Denition IV-(c), under the assumption that the forward swap rate follows the
jump-diusion process in Eq. (34). These derivations lead to the two indexes in Eq. (39) (percentage)
and Eq. (41) (basis point). Section H.1 develops the arguments applying to the percentage contract
and index, and Section H.2 contains derivations relating to the basis point.
H.1. Percentage
We apply Its lemma to Eq. (34), obtaining,
d ln1
t
(T
1
, , T
a
) =
_
E
Qswap;
_
c
)n(t)
1
_
j (t)
_
dt
1
2
|o
t
(T
1
, , T
a
)|
2
dt
o
t
(T
1
, , T
a
) d\
+
(t) ,
a
(t) d
+
(t)
=
1
2
_
|o
t
(T
1
, , T
a
)|
2
dt ,
2
a
(t) d
+
(t)
_
o
t
(T
1
, , T
a
) d\
+
(t)

_
E
Qswap;
_
c
)n(t)
1
_
j (t)
_
dt ,
a
(t) d
+
(t)
1
2
,
2
a
(t) d
+
(t) .
58 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
We have,
E
Qswap;
_
c
)n(t)
1
_
d
+
(t) =
_
E
Qswap;
_
c
)n(t)
1
_
j (t)
_
dt, (H.1)
such that by the denition of \
J
a
(t, T) in Eq. (35), and Eq. (H.1), we obtain:
2E
Qswap;
_
ln
1
T
(T
1
, , T
a
)
1
t
(T
1
, , T
a
)
_
2E
Qswap;
__
T
t
_
c
)n(t)
1 ,
a
(t)
1
2
,
2
a
(t)
_
d
+
(t)
_
= E
Qswap;
[\
a
(t, T)[
=
1
IVLI
t
(T
1
, , T
a
)
E
t
_
c

_
T
t
r
c
d:
IVLI
T
(T
1
, , T
a
) \
J
a
(t, T)
_
=
F
var,a
(t, T)
IVLI
t
(T
1
, , T
a
)
= P
J,+
var,a
(t, T) , (H.2)
where the third equality follows by the denition of the IRV forward agreement in Denition I, and
the fourth by a straightforward generalization of Eq. (10). Comparing Eq. (H.2) with Eq. (10) and
Eq. (C.2) produces Eq. (37).
H.2. Basis point
Apply Its lemma for jump-diusion processes to Eq. (34), obtaining,
d1
2
t
(T
1
, , T
a
)
1
2
t
(T
1
, , T
a
)
= 2
_
E
Qswap;
_
c
)n(t)
1
_
j (t)
_
dt 2o
t
(T
1
, , T
a
) d\
+
(t)
|o
t
(T
1
, , T
a
)|
2
dt
_
c
2)n(t)
1
_
d
+
(t)
= 2
_
E
Qswap;
_
c
)n(t)
1
_
j (t)
_
dt 2
_
c
)n(t)
1
_
d
+
(t) 2o
t
(T
1
, , T
a
) d\
+
(t)
|o
t
(T
1
, , T
a
)|
2
dt
_
c
)n(t)
1
_
2
d
+
(t) . (H.3)
By integrating, taking expectations under Q
swap
, and using the denition of basis point variance,
\
J,BP
a
(t, T) in Eq. (36), leaves:
E
Qswap;t
_
1
2
T
(T
1
, , T
a
) 1
2
t
(T
1
, , T
a
)
_
= 2
_
T
t
_
E
Qswap;
_
c
)n(t)
1
_
j (t)
_
dt 2E
Qswap;t
__
T
t
_
c
)n(t)
1
_
d
+
(t)
_
. .
=0
2E
Qswap;
__
T
t
o
t
(T
1
, , T
a
) d\
+
(t)
_
. .
=0
E
Qswap;
[\
J,BP
a
(t, T)[,
where the rst term is zero by Eq. (H.1). By Eq. (H.3), and Eq. (G.3), this cancellation leads to Eq.
(40) and, hence, Eq. (41).
59 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
References
Bakshi, Gurdip and Dilip Madan, 2000. Spanning and Derivative Security Evaluation. Jour-
nal of Financial Economics 55, 205-238.
Black, Fisher and Myron Scholes, 1973. The Pricing of Options and Corporate Liabilities.
Journal of Political Economy 81, 637-659.
Black, Fisher, 1976. The Pricing of Commodity Contracts. Journal of Financial Economics
3, 167-179.
Bossu, Sebastien, Eva Strasser and Regis Guichard, 2005. Just What You Need To Know
About Variance Swaps. JPMorgan Equity Derivatives.
Brigo, Damiano and Fabio Mercurio, 2006. Interest Rate ModelsTheory and Practice, with
Smile, Ination and Credit. Springer Verlag (2nd Edition).
Britten-Jones, Mark and Anthony Neuberger, 2000. Option Prices, Implied Price Processes
and Stochastic Volatility. Journal of Finance 55, 839-866.
Carr, Peter and Dilip Madan, 2001. Optimal Positioning in Derivative Securities. Quantita-
tive Finance 1, 19-37.
Demeter, Kresimir, Emanuel Derman, Michael Kamal and Joseph Zou, 1999. More Than
You Ever Wanted To Know About Volatility Swaps. Goldman Sachs Quantitative Strate-
gies Research Notes.
Dumas, Bernard, 1995. The Meaning of the Implicit Volatility Function in Case of Stochastic
Volatility. Available from:
[Link]
Hagan, Patrick S., 2003. Convexity Conundrums: Pricing CMS Swaps, Caps, and Floors.
Wilmott Magazine, March 38-44.
Jacod, Jean and Albert N. Shiryaev, 1987. Limit Theorems for Stochastic Processes. Springer
Verlag, Berlin.
Jiang, Rosanne, 2011. An Analysis of Gamma Trading Strategy in Swaption Market. Work-
ing Paper London School of Economics.
Mele, Antonio, 2003. Fundamental Properties of Bond Prices in Models of the Short-Term
Rate. Review of Financial Studies 16, 679-716.
Mele, Antonio, 2011. Lectures on Financial Economics. Book manuscript. Available from:
[Link]
Mercurio, Fabio and Andrea Pallavicini, 2006. Swaption Skews and Convexity Adjustments.
Banca IMI.
Neuberger, Anthony, 1994. Hedging Volatility: the Case for a New Contract. Journal of
Portfolio Management 20, 74-80.
Trolle, Anders B. and Eduardo S. Schwartz, 2011. The Swaption Cube. Working paper
EPFL and Swiss Finance Institute.
60 c _by Antonio Mele and Yoshiki Obayashi
An Interest Rate Swap Volatility Index and Contract
Vasicek, Oldrich A., 1977. An Equilibrium Characterization of the Term Structure. Journal
of Financial Economics 5, 177-188.
Veronesi, Pietro, 2010. Fixed Income Securities. John Wiley & Sons.
No statement within this paper should be construed as a recommendation to buy or sell a
nancial product or to provide investment advice. Past performance is not indicative of future
results. Options involve risk and are not suitable for all investors. Prior to buying or selling
an option, a person must receive a copy of Characteristics and Risks of Standardized Options.
Copies are available fromyour broker, by calling 1-888-OPTIONS, or fromThe Options Clearing
Corporation at [Link]. CBOE
R _
, Chicago Board Options Exchange
R _
and VIX
R _
are
registered trademarks of Chicago Board Options Exchange, Incorporated (CBOE). S&P
R _
and
S&P 500
R _
are trademarks of Standard & Poors Financial Services, LLC.
61 c _by Antonio Mele and Yoshiki Obayashi