Global Banking & Markets

November 17, 2010

Q
u
a
n
t
i
t
a
t
i
v
e
A
n
a
l
y
s
i
s
Paul Doust
Head of Client
Quantitative Analysis
+44 (0)20-7085-6015
[email protected]
Jian Chen
Quantitative Analyst
+44 (0)20-7085-0135
[email protected]
www.rbsm.com
Make it happen
TM
Using intrinsic currency
volatility to trade FX correlation
and covariance swaps
This article analyses correlation swaps and covariance swaps in terms
of the intrinsic currency volatilities of the underlying currencies. The
conclusions are that, taking USD/JPY and EUR/JPY as an example:
A covariance swap between USD/JPY and EUR/JPY which pays the
xed covariance is mostly a long position in JPY volatility;
A correlation swap between USD/JPY and EUR/JPY which pays the
xed correlation is a long position in JPY volatility coupled with
smaller short positions in USD volatility and EUR volatility.
This method of analysis can give new insight into when it makes sense
to trade these products.
1 Introduction
In the foreign exchange market, FX correlation swaps are popular
products. Covariance swaps are well known products in some markets,
and interest in FX covariance swaps is increasing. At maturity, these
products have settlements of the form
Payment = Notional (Realised - Strike) ,
where `Realised' is the realised correlation or covariance over the
lifetime of the product, and `Strike' is a xed reference correlation or
Realised correlation between USD/JPY and EUR/JPY
as USD, EUR and JPY intrinsic volatility changes
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
-6% -4% -2% 0% 2% 4% 6%
Change in intrinsic currency volatility
R
e
a
l
i
s
e
d

c
o
r
r
e
l
a
t
i
o
n
Corr vs changes in JPY vol 1 standard deviation
Corr vs changes in EURvol 1 standard deviation
Corr vs changes in USDvol 1 standard deviation
Figure 1: A rise in the intrinsic volatility of JPY causes a rise in the re-
alised correlation between USD/JPY and EUR/JPY. However, rises in
the intrinsic volatility of both USD and EUR cause falls in the realised
correlation. The dotted lines show 1 standard deviation, correponding
to possible deviation of the observed correlation from its true value.
The Royal Bank of Scotland
Q
u
a
n
t
i
t
a
t
i
v
e
A
n
a
l
y
s
i
s
|
N
o
v
e
m
b
e
r
1
7
,
2
0
1
0
covariance which is set in advance. The payment could go either way between the two counterparties of the
swap, depending on whether the realised correlation or covariance is higher or lower than the strike. Note that
the currency pairs used for these products usually have a common currency between the two FX rates involved.
The concept of intrinsic currency volatilities was rst introduced by (5), and subsequently developed in (1) and
(3). The main idea is that the foreign exchange market can be analysed in terms of the behaviour of the intrinsic
value of individual currencies, rather than FX rates which are currency pairs. In the context of FX correlation
and covariance swaps, this means that the correlations and covariances between e.g. the FX rates USD/JPY,
EUR/USD and EUR/JPY are determined by the underlying intrinsic currency volatilities of USD, EUR and JPY,
coupled with the correlations of the same currencies. One of the ingredients used when determining intrinsic
currency volatilities is a desire to keep the correlations between the different intrinsic currency values small.
The reason for this is to make the intrinsic currency values as independent of each other as possible, so that
they end up giving a good representation of the characteristics of each individual currency. The basic
mathematics of intrinsic currency values and volatilities is summarised in the appendix starting on page 6.
2 Relationship between intrinsic currency
volatilies and FX correlation and covariance swaps
It is straightforward to understand the way the covariances between FX rates change as the underlying intrinsic
currency volatilities change. Consider USD/JPY and EUR/JPY as an example. As JPY volatility gets bigger
and bigger while USD and EUR volatility remain constant, eventually the behaviour of JPY will completely
dominate the behaviour of the two FX rates. This means that for very large JPY volatility compared to USD and
EUR volatility, the correlation between USD/JPY and EUR/JPY will be close to 1. Hence as JPY volatility rises
or falls, the covariance and correlation between USD/JPY and EUR/JPY must rise and fall too.
This relationship is also evident when looking at the sensitivity of FX correlation or covariance swaps to changes
in volatility, commonly known as vega. Consider again the example of a long correlation position in USD/JPY
and EUR/JPY, and suppose that the trade size is USD 10,000 per correlation point, (i.e. per 0.01 of correlation).
Then typical vega positions in the underlying currency pairs would be as follows:
USD/JPY 22,355
EUR/JPY 49,284
EUR/USD -73,991
.
The long vega positions in USD/JPY and EUR/JPY mean that the position is long intrinsic JPY volatility twice,
and long USD and EUR volatility once each. However, to some extent the short vega position in EUR/USD will
cancel the long vegas in intrinsic USD and EUR vega. Indeed, calculating the intrinsic currency vegas from the
above currency pair vegas produces
USD -34,818
EUR -15,189
JPY 60,929
.
The magnitude of these intrinsic currency vegas corresponds to the slopes of the lines in gure 1. So as
expected, the dominant position is the long position in JPY intrinsic volatility. A similar argument was used in (4)
in connection with trading currency triangles with vanilla FX options.
A
USD
A
EUR
A
JPY
A
CAD
A
USD
1 0.26 0.01 0.06
A
EUR
0.26 1 -0.06 -0.06
A
JPY
0.01 -0.06 1 -0.33
A
CAD
0.06 -0.06 -0.33 1
Table 1: Typical correlations between the intrinsic currency values A
USD
, A
EUR
, A
JPY
and A
CAD
.
2
The Royal Bank of Scotland
Q
u
a
n
t
i
t
a
t
i
v
e
A
n
a
l
y
s
i
s
|
N
o
v
e
m
b
e
r
1
7
,
2
0
1
0
Intrinsic currency volatility
o
USD
8.0%
o
EUR
7.6%
o
JPY
12.2%
o
CAD
8.0%
Table 2: Typical intrinsic currency volatilities o
USD
, o
EUR
, o
JPY
and o
CAD
.
The formulas for the covariance and correlation between FX rates in terms of intrinsic currency variables are
given at the end of the appendix. However, instead of studying the formulas, the examples presented below and
shown in graphs in gures 1-4 can be used to understand how correlation swaps and covariance swaps behave
as the underlying intrinsic currency volatilities change. The data for the intrinsic currency volatilities and
correlations which was used for these examples is shown in tables 1 and 2. Table 1 shows the correlations
between the intrinsic currency values, and table 2 shows the intrinsic currency volatilities. Two cases will be
considered, namely the correlation/covariance between USD/CAD and EUR/CAD, and the
correlation/covariance between USD/JPY and EUR/JPY.
One issue with correlation and covariance swaps is that the measurement of correlation and covariance is
subject to sampling noise. For example, with a three month correlation swap with daily xings where the true
correlation is 40%, the one standard deviation range for correlation measurements could be 30%-50%. The
range from minus one standard deviation to plus one standard deviation covers 68% of the distribution, so 32%
of the time the measured correlation will lie outside the range 30%-50%, even though the true correlation is
40%. However, the more observations there are the smaller this effect is, so the examples used here are 1 year
correlation/covariance swaps. On all four graphs, the dotted lines show the plus/minus one standard deviation
range.
Another issue is the fact that the correlation between FX rates depends on the correlations between intrinsic
currency values, as well as the intrinsic currency volatilities. However, as mentioned in the last paragraph in the
introduction, the intrinsic currency framework often produces consistently low correlations between different
currencies. Hence for the purpose of this analysis, the effect of changing correlations on realised correlation
and covariance will be ignored. Looking at the data in table 1, an example of three currencies with low
Realised covariance between USD/CAD and EUR/CAD
as USD, EUR and CAD intrinsic volatility changes
-0.5%
0.0%
0.5%
1.0%
1.5%
2.0%
2.5%
-6% -4% -2% 0% 2% 4% 6%
Change in intrinsic currency volatility
R
e
a
l
i
s
e
d

c
o
r
r
e
l
a
t
i
o
n
Cov vs changes in CAD vol 1 standard deviation
Cov vs changes in EUR vol 1 standard deviation
Cov vs changes in USD vol 1 standard deviation
Figure 2: Table 1 shows that the correlations between USD, EUR, and CAD are all low. In this case, the
realised covariance between any two currency pairs, for example USD/CAD and EUR/CAD, mostly depends on
the intrinsic currency volatility of the common currency. Figure 3 considers the same two currency pairs and
shows how their correlation varies for changes in intrinsic currency volatility.
3
The Royal Bank of Scotland
Q
u
a
n
t
i
t
a
t
i
v
e
A
n
a
l
y
s
i
s
|
N
o
v
e
m
b
e
r
1
7
,
2
0
1
0
Realised correlation between USD/CAD and EUR/CAD
as USD, EUR and CAD intrinsic volatility changes
-20%
-10%
0%
10%
20%
30%
40%
50%
60%
70%
80%
-6% -4% -2% 0% 2% 4% 6%
Change in intrinsic currency volatility
R
e
a
l
i
s
e
d

c
o
r
r
e
l
a
t
i
o
n
Corr vs changes in CAD vol 1 standard deviation
Corr vs changes in EUR vol 1 standard deviation
Corr vs changes in USD vol 1 standard deviation
Figure 3: Correlation between forex rates with a common currency rises as the intrinsic currency volatility of
the common currency rises, but falls as the intrinsic currency of the other two currencies rise. This gure uses
USD/CAD and EUR/CAD to illustrate the behaviour. The case of USD/JPY and EUR/JPY is shown in gure
1, where the result is qualitatively the same. Figure 2 considers USD/CAD and EUR/CAD and shows how their
covariance varies for changes in intrinsic currency volatility.
Realised covariance between USD/JPY and EUR/JPY
as USD, EUR and JPY intrinsic volatility changes
0.0%
0.5%
1.0%
1.5%
2.0%
2.5%
3.0%
-6% -4% -2% 0% 2% 4% 6%
Change in intrinsic currency volatility
R
e
a
l
i
s
e
d

c
o
r
r
e
l
a
t
i
o
n
Cov vs changes in JPY vol 1 standard deviation
Cov vs changes in EUR vol 1 standard deviation
Cov vs changes in USD vol 1 standard deviation
Figure 4: A covariance swap between USD/JPY and EUR/JPY is also mostly a position in the intrinsic currency
volatility of the common currency, which is JPY in this case. This is qualitatively the same behaviour that was
observed for USD/CAD and EUR/CAD, which was shown in gure 2. However, because the correlation between
USD and JPY is not negligible, in this case there is a small dependence on the other two intrinsic currency
volatilities.
4
The Royal Bank of Scotland
Q
u
a
n
t
i
t
a
t
i
v
e
A
n
a
l
y
s
i
s
|
N
o
v
e
m
b
e
r
1
7
,
2
0
1
0
correlations is USD, EUR and CAD. Figures 2 and 3 show how the covariance and correlation between
USD/CAD and EUR/CAD change as intrinsic currency volatilities change from the values shown in table 2.
Figure 2 illustrates the fact that since the correlations between USD, EUR and CAD are low, a covariance swap
between USD/CAD and EUR/CAD is mostly a pure position in the intrinsic currency volatility of the common
currency, i.e. CAD.
With USD, EUR and JPY, however, the USD-JPY correlation in table 1 is not negligible. Figure 4 shows that in
this case, although the major sensitivity of the covariance between USD/JPY and EUR/JPY is still the intrinsic
currency volatility of the common currency (i.e. JPY), the intrinsic currency volatility of USD has also some
effect.
The dependence of correlation on changes in intrinsic currency volatilities is illustrated in gures 1 and 3. In
both cases, the major effect is that correlation rises as the intrinsic currency volatility of the common currency
rises, and to a lesser extent the correlation falls as the intrinsic currency volatilities of the other two currencies
rise.
3 Conclusion
This article has demonstrated that for covariance and correlation swaps between FX rates with a common
currency:
A covariance swap which pays the xed covariance is mostly a long position in the intrinsic currency volatility
of the common currency;
A correlation swap which pays the xed correlation is a long position in the intrinsic currency volatility of the
common currency, coupled with smaller short positions in the intrinsic currency volatility of the other two
currencies.
Hence the concept of intrinsic currency volatilities can give new insight into when it makes sense to trade these
products.
5
The Royal Bank of Scotland
Q
u
a
n
t
i
t
a
t
i
v
e
A
n
a
l
y
s
i
s
|
N
o
v
e
m
b
e
r
1
7
,
2
0
1
0
Appendix
The mathematics of intrinsic currency values
The basic concept is to model FX rates A
ij
between currency i and currency , as ratios of the intrinsic currency
values A
i
and A
j
, so that
A
ij
=
A
i
A
j
. (1)
The idea is that A
i
and A
j
somehow represent the value of the individual currencies i and ,. Note that it is not
necessary to specify the units of the A
i
because they cancel out in the ratio
1
. Although the A
i
and their
volatilities o
i
are not directly observable, various methods can be used to distill information about them from the
quantities that are observable in the FX market (see e.g. (2)). The convention where each A
i
increases when
the currency it represents becomes more valuable is adopted, which means that A
ij
is the number of units of
currency , corresponding to one unit of currency i.
The original model in (5) assumed that the A
i
followed log-normal processes, i.e.
dA
i
A
i
= j
i
dt + o
i
d\
i
, d\
i
d\
j
= j
ij
dt (2)
where o
i
and the correlations j
ij
were assumed to be constant. The more recent work in (3) extends this into a
stochastic volatility model by assuming that o
i
also follows a log-normal stochastic process. This results in a
SABR style model, which has some similarities to the original SABR model developed by (6) for the interest rate
markets.
For both the original pure-log-normal model and the newer stochastic volatility model, the instantaneous
volatility o
ij
of FX rates A
ij
is given by
o
ij
=
q
o
2
i
2j
ij
o
i
o
j
+ o
2
j
. (3)
Furthermore, given (1) the stochastic process for A
ij
is
dA
ij
A
ij
=

c
i
c
j
j
ij
o
i
o
j
+ o
2
j

dt + o
i
d\
i
o
j
d\
j
. (4)
This means that the instantaneous covariance between A
ik
and A
jk
is given by
Covariance (A
ik
, A
jk
) = o
2
k
+ j
ij
o
i
o
j
j
ik
o
i
o
k
j
jk
o
j
o
k
, (5)
and that the instantaneous correlation between A
ik
and A
jk
is given by
Correlation (A
ik
, A
jk
) =
o
2
k
+ j
ij
o
i
o
j
j
ik
o
i
o
k
j
jk
o
j
o
k
p
o
2
i
2j
ik
o
i
o
k
+ o
2
k
q
o
2
j
2j
jk
o
j
o
k
+ o
2
k
. (6)
As discussed in the introduction, in practice the correlations j
ij
, j
ik
and j
jk
are usually small, so the leading
order behaviour is usually given by:
Covariance (A
ik
, A
jk
) o
2
k
, Correlation (A
ik
, A
jk
)
o
2
k
p
o
2
i
+ o
2
k
q
o
2
j
+ o
2
k
. (7)
1
However, it is possible to understand the X
i
as measuring currency values in `goods', i.e. one can imagine there exists a universal basket
of goods dened implicitly by the forex market, which can be used to measure the value of currencies all over the world (1).
6
The Royal Bank of Scotland
References
[1] Jian Chen and Paul Doust. Estimating intrinsic currency values. RISK magazine, July:8995, 2008.
[2] Jian Chen and Paul Doust. Improving intrinsic currency analysis: using information entropy and beyond.
RBS research article, 12th November, 2009.
[3] P. Doust and J. Chen. The stochastic intrinsic currency volatility framework: A consistent model of multiple
forex rates and their volatilities. RBS research article, 22nd April, 2010.
[4] Paul Doust. Currency triangle trading strategies in the forex option market. RBS research article, 19th June,
2007.
[5] Paul Doust. The intrinsic currency valuation framework. RISK magazine, March:7681, 2007.
[6] P. Hagan, D. Kumar, A. Lesniewski, and D. Woodward. Managing smile risk. Wilmott Magazine, pages
84108, July 2002.
This material has been prepared by The Royal Bank of Scotland plc (RBS) for information purposes only and is not an offer to buy or sell or a solicitation of an offer to buy or sell any security or instrument or
to participate in any particular trading strategy. This material should be regarded as a marketing communication and may have been produced in conjunction with the RBS trading desks that trade as principal in
the instruments mentioned herein. This commentary is therefore not independent from the proprietary interests of RBS, which may conict with your interests. Opinions expressed may differ from the opinions
expressed by other divisions of RBS including our investment research department. This material includes analyses of securities and related derivatives that the rms trading desk may make a market in,
and in which it is likely as principal to have a long or short position at any time, including possibly a position that was accumulated on the basis of this analysis prior to its dissemination. Trading desks may
also have or take positions inconsistent with this material. This material may have been made available to other clients of RBS before being made available to you. Issuers mentioned in this material may be
investment banking clients of RBS. Pursuant to this relationship, RBS may have provided in the past, and may provide in the future, nancing, advice, and securitization and underwriting services to these clients in
connection with which it has received or will receive compensation. The author does not undertake any obligation to update this material. This material is current as of the indicated date. This material is prepared
from publicly available information believed to be reliable, but RBS makes no representations as to its accuracy or completeness. Additional information is available upon request. You should make your own
independent evaluation of the relevance and adequacy of the information contained in this material and make such other investigations as you deem necessary, including obtaining independent nancial advice,
before participating in any transaction in respect of the securities referred to in this material.
THIS MATERIAL IS NOT INVESTMENT RESEARCH AS DEFINED BY THE FINANCIAL SERVICES AUTHORITY.
United Kingdom. Unless otherwise stated herein, this material is distributed by The Royal Bank of Scotland plc (RBS) Registered Ofce: 36 St Andrew Square, Edinburgh EH2 2YB. Company No. 90312.
RBS is authorised and regulated as a bank and for the conduct of investment business in the United Kingdom by the Financial Services Authority. Australia. This material is distributed in Australia to wholesale
investors only by The Royal Bank of Scotland plc (Australia branch), (ABN 30 101 464 528), Level 48 Australia Square Tower, 264278 George Street, Sydney NSW 2000, Australia which is authorised and
regulated by the Australian Securities and Investments Commission, (AFS License No 241114), and the Australian Prudential Regulation Authority. France. This material is distributed in the Republic of France
by The Royal Bank of Scotland plc (Paris branch), 94 boulevard Haussmann, 75008 Paris, France. Hong Kong. This material is being distributed in Hong Kong by The Royal Bank of Scotland plc (Hong Kong
branch), 30/F AIG Tower, 1 Connaught Road, Central, Hong Kong, which is regulated by the Hong Kong Monetary Authority. Italy. Persons receiving this material in Italy requiring additional information or wishing
to effect transactions in any relevant Investments should contact The Royal Bank of Scotland plc (Milan branch), Via Turati 18, 20121, Milan, Italy. Japan. This material is distributed in Japan by The Royal Bank of
Scotland plc (Tokyo branch), Shin-Marunouchi Center Building 19F21F, 6-2 Marunouchi 1-chome, Chiyoda-ku, Tokyo 100-0005, Japan, which is regulated by the Financial Services Agency of Japan. Singapore.
This material is distributed in Singapore by The Royal Bank of Scotland plc (Singapore branch), 1 George Street, #10-00 Singapore 049145, which is regulated by the Monetary Authority of Singapore. RBS is
exempt from licensing in respect of all nancial advisory services under the (Singapore) Financial Advisers Act, Chapter 110.
United States of America. RBS is regulated in the US by the New York State Banking Department and the Federal Reserve Board. The nancial instruments described in the document comply with an applicable
exemption from the registration requirements of the US Securities Act 1933. This material is only being made available to U.S. persons that are also Major U.S. institutional investors as dened in Rule 15a-6 of the
Securities Exchange Act 1934 and the interpretative guidance promulgated thereunder. Major U.S. institutional investors should contact Greenwich Capital Markets, Inc., (RBS Greenwich Capital), an afliate of
RBS and member of the NASD, if they wish to effect a transaction in any Securities mentioned herein.