Currency benchmarks for

investors
by Jason Batt and Bilal Hafeez, Deutsche Bank

End users have embraced the Deutsche Bank Currency Indices

because they provide a cheap and efficient means of gaining
exposure to the returns inherent in the FX market, they are very
liquid and transparent, they allow non-specialist FX market
participants to benefit from Deutsche Banks (DB) intellectual
capital and leading market position, and exposure to them is
offered in a number of simple and easy-to-trade formats.

Whether it is competitive pressures from other fund managers, The real paradigm shift has been to recognise these
or stakeholder pressure to meet increasing pension fund returns as beta rather than alpha alone; in other words,
liabilities, investors are continually being pushed to generate to acknowledge that there is an inherent return available
higher returns with less volatility. One of the best ways to in currency markets. Importantly, we can show that

achieve this is to include more assets that simultaneously

offer higher returns and increased diversication. It therefore
comes as no surprise that investments other than bonds and
equities have proven to be an area of great focus in recent
years. The biggest pitfall of many investments outside of
bonds and equities, however, is their relative lack of liquidity,
which makes larger allocations prohibitive. In this context,
foreign exchange stands out. It has the highest daily turnover
Jason Batt Bilal Hafeez
of any market, a track record of positive excess returns and it
offers true diversication to bonds and equities.
Jason Batt, Global Head of Currency Index Products
In a March 2007 DB Research piece: Currency Markets: Is tel: +44 (20) 7547 1661
money left on the table? we showed that currency markets e-mail: jason.batt@db.com
have inherent positive returns over time. The existence of
a sizeable proportion of market participants who do not Bilal Hafeez, Managing Director and
maximise profits, the fact that market participants have Global Head of FX Strategy
different objectives and beliefs, and the historically positive tel: +44 (20) 7545 0354
returns of following three investment approaches: carry, e-mail: bilal.hafeez@db.com
momentum and valuation all provide evidence for this.

The Euromoney Foreign Exchange & Treasury Management Handbook 2008

benchmarks in other asset classes that are viewed as Positive momentum similar to market cap weighting
representing the market return, or beta, are in fact trading and minimum capitalisation rules.
rules that capture the bulk of returns of the given market. The 10 currencies in the DBCR Index universe are
ranked each month by their annual return against the

Currencies: last one to join the US$. The DBCR Momentum Index goes long the three
currencies with the highest return against the US$
benchmark club?
and short the three currencies with the lowest return
The absence of a benchmark for the currency market is
against the US$.
quite conspicuous, given that the daily turnover now
Undervalued similar to incorporating a fundamental
exceeds US$2 trillion and the market has been trading
metric such as earnings or revenue as often used in
for more than 20 years since the end of Bretton Woods in
equity indices.
1973 and the more widespread adoption of capital account
For valuation we rank the 10 currencies according to
convertibility through the late 1970s and early 1980s.
their purchasing power parity (PPP) exchange rates
When viewed as a set of trading rules, the accepted
and go short the three with the highest spot-exchange
benchmarks of other asset classes indicate a level of
rate relative to the fair value level of PPP, and go long
subjectivity that would not otherwise be apparent. In fact,
the three with the lowest spot-exchange rate relative
they really reflect a set of transparent rules that capture
to the fair value level of PPP. In this way, though we
a substantial portion of the returns of a given market. By
may arrive at lower returns, we get around the issue
being widely followed, they become benchmarks.
of picking arbitrary over- and undervaluation extremes
such as +/-20%, which could lead to a risk of ex-post
Currency beta: the Deutsche Bank optimisation
Currency Returns (DBCR) Index Finally, we create our benchmark DBCR Index by simply
DB has identified four broad rules by which to take taking the average of the three strategies (rules).
positions in currencies, such that the characteristics of
these rules are consistent with those of other Brief summary of the rationale for
established benchmarks. these rules
Select from the currencies of the Developed World to Carry
meet liquidity requirements. Exploits the widely observed forward premium puzzle
The DBCR Index is constructed from the currencies of the or forward rate bias, which suggests that systematically
G10 economies: the US dollar (US$), euro (), Japanese buying high interest rate currencies and selling low
yen (), British pound (), Swiss franc (SFr), Norwegian interest rate currencies may be profitable. This is because
krone (NKr), Swedish krona (SKr), Australian dollar (A$), of the existence of a risk premia, the use of different
New Zealand dollar (NZ$) and Canadian dollar (C$). models to forecast currencies by rational market

Positive net yield (or carry) similar to bond indices. participants and the differing constraints and objectives
faced by market participants.
The 10 currencies in the DBCR Index universe are
ranked each quarter according to their three-month Momentum
LIBOR yield. The DBCR carry index goes long the three Currencies appear to trend over time, which suggests
currencies with the highest yields and short the three that using past prices may be informative to investing in
currencies with the lowest yields. currencies. This is due to the existence of irrational traders,

The Euromoney Foreign Exchange & Treasury Management Handbook 2008

the possibility that prices provide information about A note on optimisation in indices
non-fundamental currency determinants or that prices may
adjust slowly to new information. Modern finance theory was one cause of last
summers turmoil
Valuation Although August 2007 may seem a long time ago, the issues
In the long run, currencies tend to move back to their fair
of how central banks should respond to such events, the
value based on (PPP). However, in the short- to medium-run,
complexity of nancial products and the disintermediation
currencies can deviate considerably from their PPP values
of banks from the credit process have been noted. Yet the
due to trading, information and other costs. This allows for
biggest lesson has not been fully appreciated. That is, our
the possibility of profiting from currencies as they revert
commonly used notions of risk must be wrong if hedge funds
back to their fair values over the long-run.
are suffering one-in-100,000 year losses, the market from
which the risk-free rate is taken (money markets) suddenly
Easy to pick the rules, but it needs becomes the riskiest market, and once-in-a-lifetime events

to be investable recur (remember LTCM in 1998, and the stock market crash
of 1987). The culprit is a dogmatic belief in modern nance
DBs investable indices are designed with regard to some
theory that results in the under-appreciation of true risk
quite obvious considerations such as liquidity, availability
and over-investment in certain markets or strategies. We can
of reliable market data, the frequency of rebalancing and
see this clearly in currency markets in general, and the carry
the sources of fixing references; these are particularly
trade in particular.
relevant for the FX market, which trades on an OTC basis
rather than on an exchange where closing levels could The paradox of optimisation and carry
reliably be used. There are also some more esoteric factors An FX carry trade takes place when an investor borrows
that are a critical part of the intellectual property that the in a low-interest rate currency, such as the Japanese yen,
non-specialist investor is leveraging when they invest in a and invests in a high-interest rate currency, such as the
DB Currency Index. New Zealand dollar. It is common to construct a portfolio
of carry trades to diversify the risk. Such portfolios have
It needs to be reliable
delivered strong returns over the decades, comparable to
Another very important factor is ensuring the robustness
those in equity markets. To further enhance these returns,
of index performance over time. DB indices are built in
one of the central tools of modern nance theory is used;
a logical order which ensures their robustness. We start
mean-variance optimisation (MVO). This technique adjusts
with trading logic which is corroborated by research and
the weights of each currency pair in the portfolio to deliver
economic rationale. A set of trading rules are built from this
the highest returns for a given level of volatility. The process
which forms the rules governing the index construction and
by which this is done is surprisingly simple. All we need are
calculation. These rules are then tested over a lengthy data
the expected returns, volatilities and correlations of the
set. This order of construction ensures that results will be
currency pairs in the portfolio. We can then optimise the
robust over time across multiple data sets, avoiding the out
portfolio for the ideal degree of diversication.
of sample bias that many optimised indices have because
they were constructed from data mining rather than logical This is all well and good, however we must realise that
trading rules. This last point is often overlooked but investors MVO is built on the idea that markets are efficient; that is,
should be particularly aware that an over-complicated set investors are just as likely to experience losses as gains. It
of index rules can have parameters optimised in such a way may seem like a technical point, but critical in this context
that it is unrealistic to assume historic performance will be because FX carry trades seem to be profitable precisely
repeated after the index launch date. because markets are not efficient. If they were, even in the

The Euromoney Foreign Exchange & Treasury Management Handbook 2008

presence of acceptable levels of risk premia, then FX carry an annualised excess return of 4.6% with a Sharpe ratio of
trades would not be profitable. The paradox then is that 0.78. It appears that returns broadly follow a cycle so, as in
investors are employing an investment strategy premised equities, currencies appear to have bull-and-bear markets,
on markets not being efficient, while at the same time with bull markets tending to last much longer than bear
employing an optimisation technique that is premised on markets (see Exhibit 1).
markets being efficient. But you cannot have it both ways.
In comparison to the DBCR, the investment return statistics
for the component strategies are lower for momentum and
The nuisance of extreme returns
valuation, but in some cases are higher for carry (Exhibit
The aws are more apparent if we look at the assumptions
1). However, the DBCR draw-downs are unambiguously
behind MVO, namely, that returns are normally distributed,
lower than any of the individual strategies and the Sharpe
and volatilities and correlations are stable. If FX carry returns
ratio is higher over the long-run. This is due to the fact that
were normally distributed then extreme gains or losses would
strategies tend to perform well at different times, and so
not have much impact on overall returns. Yet, if we exclude
the correlations are low and often negative between the
the 25 largest one-day losses since 1980, we nd that FX carry
strategies (Exhibit 1). It appears that valuation and carry
returns would be dramatically higher; US$100 invested in
have positive correlations, while momentum has a negative
1980 would have grown to US$6000, rather than U$3000, if all
correlation with both.
returns are included. From another angle, in a normal world,
a one-day loss of more than 3% would only be seen once every The low correlations between the individual strategies
4,000,000 years. Since 1980, weve already seen four such manifest themselves in a lowered overall DBCR volatility.
extreme trading days! The fundamental aw in MVO is that it This reduces mark-to-market volatility for investors who
optimises to average risk, when in fact we should care more are reporting earnings more frequently and importantly it
about extreme downside risk. reduces the volatility and hence the cost of hard options
bought on the DBCR Index, allowing for increased leverage
The second assumption is as shaky; both volatilities and
when investing via options.
correlations are unstable in the real world. Focusing on NZ$/
volatilities, we nd that one-year (historic) volatility has ranged
from 5% to 29% with even wider swings for one-month volatility. What about alternative benchmarks?
Worryingly, periods of low volatility tend to get interrupted by
Just as multiple benchmarks exist in equity and fixed
sharp spikes in volatility triggered by carry trade losses. Using
income markets, DB believes that multiple benchmarks are
MVO, this has the unfortunate implication of giving signals to
also appropriate for FX markets.
increase ones exposure to carry trades just before sharp
losses, and then reducing ones exposure after them, which The three sub-indices of the DBCR broad market benchmark
prevents capturing the typical recovery in prots. Correlation lend themselves to becoming benchmarks for their
is no different. During periods of carry losses, correlations respective strategies. DB currently makes markets in
between currency pairs pick up and so the diversication one all three DBCR sub-indices in various formats including
thought one had fails to hold when most needed. contracts for difference and options.

Of particular interest to the investor community has been

How does the Deutsche Bank Currency DBs harvest family of indices.

Returns (DBCR) Index perform? DB Currency Harvest

Since 1980, the DBCR has delivered excess returns of close For over two years DB has been marketing variations on
to 4% with a Sharpe ratio of 0.80 and a maximum peak-to- dynamic carry indices. Dynamic indices perform much
trough drawdown of 11%. Between 2002 and 2007, it had better over time than static carry indices. Although there

The Euromoney Foreign Exchange & Treasury Management Handbook 2008

 DBCR returns Exhibit 1

Excess returns of DBCR Rolling 2-year risk-adjusted returns for DBCR

25 3.0

2.5
20
Excess returns (over risk-free rate),
includes transactions costs %

2.0

2-year rolling Sharpe ratio

15
1.5

10 1.0

0.5
5
0.0

0 -0.5

-1.0
-5
80 82 84 86 88 90 92 94 96 98 00 02 04 06 82 84 86 88 90 92 94 96 98 00 02 04 06

Summary statistics of DBCR and components 1980-2006 1990-2006 2000-2006

DBCR
Excess returns* 4.0% 3.9% 4.9%
Volatility 5.2% 5.1% 5.3%
Sharpe ratio 0.77 0.76 0.94

Maximum drawdown -11%

Carry 4.9% 5.1% 7.0%

0.59 0.65 0.98

Momentum 3.0% 2.8% 3.5%

0.35 0.32 0.46

Valuation 4.1% 3.8% 4.3%

0.46 0.41 0.54

* Includes transaction costs and carry, and excludes legacy euro currencies, save DEM. Including them, would have
kept returns close-to-unchanged for DBCR

Source: DB Global Markets Research

The Euromoney Foreign Exchange & Treasury Management Handbook 2008

are some core components to the carry trade such as being
A summary of the three indices returns
short against long NZ$, many currencies, such as the Exhibit 2
and Sharpe ratios for the period 2002 to 2007
US$, NKr and SKr, have spent time as both a long and short
currency in the highest yielding carry basket, while others
Annualised Sharpe
have been selected as a long or short at various times and
Index excess return ratio
been omitted from the component currencies at others.
DB Balanced
A diversied carry basket is much more efcient than merely Currency Harvest 15.15% 1.67
selecting the one highest-yielding currency to be long of and
the one lowest yielding currency to be short of. A basket of DB Global
Currency Harvest 12.03% 1.35
three to ve currencies to be long and short of still captures
the bulk of the carry differential in the currencies yields, but DB G10
has a much lower volatility than a single currency pair. Currency Harvest 6.84% 0.87

In addition to the G10 Harvest Index, which forms the carry

component of the DBCR benchmark Index, there are two
dynamic carry indices which draw on currencies for an expanded Source: DB Global Markets Research

universe of twenty currencies, the DB Balanced Currency

Harvest Index and the DB Global Currency Harvest Index. Global Currency Harvest also selects the ve highest yielding

In addition to the 10 G10 currencies, 10 other currencies may currencies to be long of and the ve lowest yielding currencies
be included in the Balanced and Global Harvest Indices, to be short of, with selections coming from the entire universe
drawn from the Korean won (W-offshore), Singaporean dollar of twenty currencies, but there are no restrictions on which
(S$-offshore), Taiwanese dollar (NT$ offshore), Mexican currencies may be selected as high or low yielders.
peso (Mex$), Brazilian real (R$ offshore), South African
Typically, the non-G10 currencies have the extremes of
rand (R), Polish zloty (Zl), Czech koruna (Kc), Hungarian forint
interest rates and hence would be selected as the long
(Ft), and the Turkish lira (TL).
or short components of each index on each rebalancing
The three Harvest Indices are constructed in broadly the date. The main risk in these types of strategy is the
same way. They each select which currencies to go long exchange rate, and in the event that there are shocks to
and short of based on three-month LIBORs and they each one non-G10 currency, contagion can become an issue as
rebalance on a quarterly basis, utilising the innovative roll many non-G10 currencies move in the same way. For this
window mechanism which helps ensure returns from these reason, the DB Balanced Currency Harvest Index has been
indices stay with those who are invested in them. the most popular of the dynamic carry indices because it

The major difference between these indices is the rule on forces a generally more diversified selection of currencies,

composition. The G10 Currency Harvest Index selects the three with a balance of G10 and non-G10 exposures. It has also
highest yielding currencies to be long and the three lowest historically exhibited the best Sharpe ratio.
yielding currencies to be short, with the selections limited
to the G10 currencies. The Balanced Currency Harvest Index
In summary
selects the ve highest yielding currencies to be long and the
ve lowest yielding currencies to be short, with the selections DB has created the Deutsche Bank Currency Returns (DBCR)
coming from the entire universe of 20 currencies, but with the Index to act as a practical broad benchmark for returns
caveat that at least two of the high yielding currencies and inherent in the currency market, in addition to creating
two of the low yielding currencies must be G10 currencies. The other more narrow benchmarks for specific strategies.

The Euromoney Foreign Exchange & Treasury Management Handbook 2008

 Asset classes Exhibit 3

Excess returns across asset classes

6

4
%

3 1980-2006
1990-2006
2

0
Currencies Bonds Equities

Risk-adjusted returns across asset classes

0.90
0.80
0.70
Sharpe ratio

0.60
0.50
0.40 1980-2006
1990-2006
0.30
0.20
0.10
0
Currencies Bonds Equities

Correlation between asset classes

Bond Equity DBCR FX FX FX

carry Mom. value

Bond 100%
Equity 26% 100%
DBCR -21% 5% 100%

FX carry -16% 4% 74% 100%

FX mom. 3% -2% 38% -6% 100%
FX value -25% 7% 66% 40% -25% 100%

Source: DB Global Markets Research, Lehmans Global Aggregate Index for bonds, and MSCI World for equities

The Euromoney Foreign Exchange & Treasury Management Handbook 2008

These indices are created from simple, logical and excess returns makes these benchmark currency indices
transparent trading rules, ensuring robustness and an attractive addition to existing investment portfolios and
reliability over time, and are investable in several formats. should make them more efcient. The conclusions of DBs

In their own right, these indices have admirable return and August 2006 research piece Currencies: pensions saviour?,

risk characteristics, but their most attractive feature is the which advocated a 20%-30% allocation to currencies in a
relatively low correlation to the returns from more traditional global asset allocation context are validated by the historical
asset classes such as equity and xed income. This low return, risk and correlation prole of the benchmark indices
correlation combined with a solid history of average annual that DB has created.

<AD76AB6G@:IH < 9BE8<:A8K6;4A:8

Ejii^c\djgXa^Zcihghi^c;M
V\V^cVcYV\V^cVcYV\V^c#
8dch^hiZciYZa^kZgn#8dch^hiZci^ccdkVi^dc#
8dch^hiZciangZXd\c^hZY#
5fcibXh\Ykcf`X cifUW_bck`YX[YX`YUXYfg\]d]b:LYadckYfg
ighcdfcj]XYUWcadYh]h]jYYX[YZcfcifW`]Ybhg"CiffYgYUfW\ 
gU`Yg UbX hfUX]b[ dfcZYgg]cbU`g UfY XYX]WUhYX hc WfYUh]b[ UbX
XY`]jYf]b[ fcibX h\Y W`cW_ jU`iY" ;`cVU` cf `cWU`  cif W`]Ybhg
VYbYh Zfca dfcjYb hYW\bc`c[m UbX U fYWcfX cZ ]bbcjUh]cb
gYWcbXhcbcbY"=h]gh\Ub_ghcUdUgg]cbZcfdihh]b[cifW`]Ybhg
fghh\UhkYWcbh]biYhcdYfZcfak]h\YlWY``YbWY]b:L"
lll#YW#Xdb

H\]gUXjYfh]gYaYbh\UgVYYbUddfcjYXUbX#cfWcaaib]WUhYXVm8YihgW\Y6Ub_5;@cbXcb"H\YgYfj]WYgXYgWf]VYX]bh\]gUXjYfh]gYaYbhUfYdfcj]XYXVm8YihgW\Y6Ub_5;cfVm]hg
giVg]X]Uf]YgUbX#cfUZZ]`]UhYg]bUWWcfXUbWYk]h\Uddfcdf]UhY`cWU``Y[]g`Uh]cbUbXfY[i`Uh]cb"8YihgW\Y6Ub_5;]gUih\cf]gYXibXYf;YfaUb6Ub_]b[@UkWcadYhYbhUih\cf]hm.6U:]b
:YXYfU`:]bUbW]U`GidYfj]g]b[5ih\cf]hmUbXk]h\fYgdYWhhcI?WcaacX]hmXYf]jUh]jYgVig]bYggVmh\Y:]bUbW]U`GYfj]WYg5ih\cf]hm/fY[i`UhYXVmh\Y:]bUbW]U`GYfj]WYg5ih\cf]hmZcf
h\YWcbXiWhcZI?Vig]bYgg"GYWif]h]YgUbX]bjYghaYbhVUb_]b[UWh]j]h]Yg]bh\YIb]hYXGhUhYgUfYdYfZcfaYXVm8YihgW\Y6Ub_GYWif]h]Yg=bW" aYaVYfBMG9 B5G8UbXG=D7 UbX]hg
Vfc_Yf!XYU`YfUZZ]`]UhYg"@YbX]b[UbXch\YfWcaaYfW]U`VUb_]b[UWh]j]h]Yg]bh\YIb]hYXGhUhYgUfYdYfZcfaYXVm8YihgW\Y6Ub_5; UbX]hgVUb_]b[UZZ]`]UhYg"7cdmf][\h8YihgW\Y6Ub_&$$,"