The Deutsche Bank Guide To Exchange Rate Determination FX Broker DB

User Manual:

Open the PDF directly: View PDF PDF.
Page Count: 160 [warning: Documents this large are best viewed by clicking the View PDF Link!]

DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 1
Deutsche Bank@
May 2002
FX Research
Global Markets Research
Michael R. Rosenberg
Head of Global FX Research
David Folkerts-Landau
Managing Director, Head of
Global Markets Research
Exchange-Rate Determination
in the Short, Medium, & Long Run
The Deutsche Bank
Guide to
Exchange-Rate
Determination
A Survey of
Exchange-Rate
Forecasting Models and
Strategies
Bandwagon Effect/
Trend-Following Behavior
Real
Interest-Rate
Differentials
Purchasing Power Parity
Capital
Flows
Fiscal
Policy Portfolio-Balance
Considerations
Current
Account
Trends
Relative
Economic
Growth
Monetary
Policy
Investor Positioning
FX Options Market
Positioning
Risk Appetite
Investor Sentiment
Net Foreign Assets
Productivity Trends
Savings/Investment
Balance Trends
Persistent Trend in
Terms-of-Trade
Exchange Rate
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
2Deutsche Bank Foreign Exchange Research
Introduction .................................................................................................... 4
Exchange Rate Determination in the Short Run ........................................ 7
Currency Forecasting Using Technical Analysis .........................................11
Sentiment and Positioning Indicators ....................................................... 21
Currency Options Market ......................................................................... 22
Order Flow and the Determination of Exchange Rates ........................... 24
Investor Positioning and the Trend in Exchange Rates ............................. 27
Risk Appetite Shifts and Currency Trends................................................. 28
Exchange Rate Determination in the Long Run ....................................... 31
Purchasing Power Parity ........................................................................... 33
The Macroeconomic-Balance Approach to Long-Run
Exchange-Rate Determination ............................................................ 42
The Real Long-Run Equilibrium Exchange Rate ....................................... 43
Productivity Trends and Exchange Rates .................................................. 46
Terms of Trade and Exchange Rates......................................................... 50
Net International Investment and the Equilibrium Exchange Rate........... 51
Long-Term Cycles in Exchange Rates....................................................... 53
Overshooting Exchange Rates ................................................................. 60
Exchange Rate Determination in the Medium Run.................................. 63
International Parity Conditions.................................................................. 65
Real Interest-Rate Differential Model ....................................................... 67
Forward-Rate Bias Strategy...................................................................... 72
Current-Account Imbalances and the Determination of
Exchange Rates................................................................................... 77
Capital Flows and Exchange Rates........................................................... 86
Mundell-Fleming Model ........................................................................... 96
Monetary Approach ................................................................................ 104
Portfolio-Balance Approach ..................................................................... 108
Fiscal Policy .............................................................................................115
Economic Growth....................................................................................119
Central-Bank Intervention ....................................................................... 124
Anticipating Currency Crises in Emerging Markets ............................... 129
References................................................................................................... 155
Table of Contents
Datastream International, Inc. is the source of the majority of data
used in the charts and tables in this publication. Other sources are
noted individually.
Market sentiment data is by permission of Consensus, Inc.,
Con-
sensus, National Futures and Financial Weekly,
(1) (800) 383-1441 or (1) (816) 373-3700
www.consensus-inc.com
Sources:
May 5, 2002
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 3
"If you think writing about the fortunes of the stock
market is tricky, try getting your arms around cur-
rencies."
Bill Gross
PIMCO
Investment Outlook, January 2002
"Having endeavored to forecast exchange rates
for more than half a century, I have understand-
ably developed significant humility about my abil-
ity in this area...."
Alan Greenspan
Remarks Before the Euro 50 Roundtable
Washington D.C., November 30, 2001
Deutsche Bank Guide to Exchange-Rate Determination
"Explaining the yen, dollar and euro exchange rates
is still a very difficult task, even ex-post."
Kenneth Rogoff
Economic Counselor and Director of Research,
International Monetary Fund
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
4Deutsche Bank Foreign Exchange Research
Getting the exchange rate right is a critical objective of all
international investors. Unfortunately, getting the exchange
rate right on a reasonably consistent basis is far from easy.
As anyone involved in the business of currency forecast-
ing can attest, it can be a humbling experience.
Currency forecasts can go awry for a variety of reasons.
For instance, if one's expectation of the direction in which
fundamental-based forces are heading is flawed, so will
be one's forecast of a currency's future path. Even if one's
interpretation of the underlying fundamental forces were
correct, currency forecasts might still go awry if short-term
technical forces carried exchange rates far from their fun-
damental equilibrium path.
Scores of empirical studies have found that fundamental-
based models tend to perform poorly in terms of explain-
ing exchange-rate trends, particularly over short-term peri-
ods. However, fundamental-based models tend to work
better over medium and especially longer-run horizons.
Unfortunately, most fund managers, whose performances
are evaluated over relatively short time spans these days,
are often not willing to risk significant sums of capital on
the basis of longer-term, fundamental-based projections.
That is why many market participants have recently turned
their attention away from longer-run fundamental-based
forecasting approaches in favor of shorter-term forecast-
ing tools such as technical-based trend-following trading
rules. In addition, there has recently been significant inter-
est in flow, sentiment, and positioning indicators to deter-
mine the exposure of market participants to the individual
currencies. Such indicators are often used as contrarian
indicators to determine whether a currency is significantly
overbought or oversold, and thus ripe for a correction.
Given the wide variety of forecasting approaches, we
thought it would be useful to put together a guidebook
that summarized each of those approaches in an easy-to-
read format. Our intention was to create a user-friendly
format where the written text was purposely kept to a
minimum and where the charts and tables—about 400 in
all—would tell the story.
This guidebook recognizes that the tools required for short-
term investors differ significantly from those needed for
medium and long-term currency managers. Hence, the
guidebook devotes separate chapters to the determina-
tion of exchange rates over short, medium, and long-term
horizons.
The adjacent schematic diagram provides a convenient il-
lustration of the layout of this guidebook and highlights
the myriad of channels through which fundamental and
technical forces jointly affect exchange rates. Some of those
channels will tend to exert a more profound impact on ex-
change rates in the short run, while others will tend have a
more profound impact in the medium or long run. We ex-
plore each of those channels in the chapters that follow.
In the chapter entitled "Exchange Rate Determination in
the Short Run," we investigate the potential risks and re-
wards of using a variety of short-run forecasting tools in
formulating short-term FX strategies. These include mov-
ing-average trend-following trading rules, sentiment and
positioning surveys, FX dealer customer-flow data, infor-
mation embedded in currency option prices, and risk ap-
petite indices.
We find that moving-average trading rules would have gen-
erated significant risk-adjusted excess returns over rela-
tively long periods for most major currency pairs, although
losing trades tend to occur far more frequently than win-
ning trades, in many cases by a factor of 3 to 1. The high
frequency of losing trades suggests that moving-average
trading rules can be risky over short periods and that an
investor would need considerable risk capital on hand to
absorb such losses to stay in the game until exchange rates
eventually become more highly trended.
The evidence on flow, sentiment, and positioning surveys
suggests that these indicators should be viewed more as
contemporaneous rather than as leading indicators of ex-
change-rate movements. We argue that such indicators can
be useful as confirmation indicators in conjunction with
traditional trend-following trading rules in formulating short-
term FX strategies.
In the chapter entitled "Exchange Rate Determination in
the Long Run," we explore the fundamental forces that
give rise to long-term cycles in exchange rates. The chap-
ter begins by noting that deviations from estimated PPP
values have tended to be large and persistent, suggesting
that fundamental forces other than relative national infla-
tion rates have played a key role in driving the long-term
path that exchange rates have taken. We investigate a va-
riety of fundamental variables that have had some suc-
cess in explaining the long-term path that currencies have
taken, including relative productivity growth, persistent
trends in a country's terms of trade, long-term trends in
net foreign asset and liability positions, and long-term
trends in national savings and investment.
Introduction
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 5
In the chapter entitled "Exchange Rate Determination in
the Medium Run," we investigate a wide range of cyclical
forces that have caused currencies either to rise or fall on
a medium-term basis relative to their long-run equilibrium
path. In many cases, these medium-term deviations from
the long-run equilibrium path have been quite sizeable and
persistent. We find that medium-term trends are influenced
by a variety of macroeconomic indicators such as the trend
in real interest-rate differentials, current and capital-account
balances, relative monetary and fiscal policies, relative eco-
nomic growth, and portfolio-balance considerations.
Finally, in the chapter entitled "Anticipating Currency Cri-
ses in Emerging Markets," we set out to identify those
economic and financial variables that have had success in
correctly predicting whether an emerging-market currency
might be vulnerable to a speculative attack, and whether it
is possible to construct an early-warning system that can
successfully pinpoint when a speculative attack might oc-
cur. Empirical research finds that crises-prone currencies
typically display a number of classic symptoms that warn
of an impending attack. Those symptoms include exces-
sive real appreciation of the emerging-market currency,
weak domestic economic growth, rising unemployment,
an adverse terms of trade shock, deteriorating current-ac-
count balances, excessive domestic credit expansion, bank-
ing-system difficulties, unsustainably large government
budget deficits, overly expansionary monetary policies, a
high ratio of M2 money supply to reserves, foreign-ex-
change reserve losses, falling asset prices, and/or a huge
build-up in short-term liabilities by either the private or public
sector.
The overall conclusion one draws from a variety of empiri-
cal studies is that the success of early-warning systems in
terms of generating correct out-of-sample projections is
mixed. While most early-warning models can point to a
significant number of correctly predicted crises, those
same models also have a tendency to generate a sizable
number of false alarms and missed crises.
Perhaps all that one can say after reviewing all the differ-
ent approaches to exchange-rate determination is that no
single approach has a monopoly on being right all of the
time. Some strategy systems such as moving-average trad-
ing rules and forward-rate bias strategies appear to have a
long-run track record of success, but one needs to be mind-
ful that there were a number of periods in the past when
significant losses were incurred from following these strat-
egies. Likewise, some key fundamental variables may have
closely tracked the trend in exchange rates in the past, but
there is no guarantee that they will continue to do so in
the future. If divergent trends begin to set in, fund manag-
ers must decide whether to disregard the trend in those
key fundamental variables or not.
Although many fund managers might prefer to follow a
rigid rule or trading system for formulating FX strategies,
one should not sell short a judgment-based approach to
currency investment management. In the end, successful
FX management is based as much on "art" as it is on "sci-
ence."
Bandwagon Effect/
Trend-Following Behavior
Real
Interest-Rate
Differentials
Purchasing Power Parity
Capital
Flows
Fiscal
Policy Portfolio-Balance
Considerations
Current
Account
Trends
Relative
Economic
Growth
Monetary
Policy
Investor Positioning
FX Options Market
Positioning
Risk Appetite
Investor Sentiment
Net Foreign Assets
Productivity Trends
Savings/Investment
Balance Trends
Exchange-Rate Determination
in the Short, Medium, and Long Run
Persistent Trend in
Terms-of-Trade
Exchange Rate
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
6Deutsche Bank Foreign Exchange Research
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 7
A Stylized Model of Exchange-Rate Overshooting
at the End of a Long-Term Uptrend
Currency’s
Value
Time
Economic Fundamentals
Exchange-Rate Determination
in the Short Run
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
8Deutsche Bank Foreign Exchange Research
Economists have come up with a wide range of theories
to explain how exchange rates are determined. The over-
whelming body of evidence from scores of empirical stud-
ies indicates that fundamental-based models, while use-
ful in explaining long-term trends, have not met much suc-
cess in explaining short-term exchange-rate trends. Indeed,
the evidence suggests that for short-term horizons, a ran-
dom walk characterizes exchange-rate movements better
than most conventional fundamental-based exchange rate
models.
One of the reasons why researchers have not been able
to unearth any significant relationship between changes
in macroeconomic variables and changes in exchange rates
over short periods is that exchange rates often exhibit much
greater variability than do macroeconomic time series in
the short run. The often chaotic behavior of exchange rates
is capable of generating so much noise that it may ob-
scure any discernable relationship between macroeco-
nomic time series and the short-term movement of ex-
change rates.
Bandwagon effects are also capable of causing exchange
rates to wander away from fundamental equilibrium val-
ues in the short run. Survey studies find that FX market
participants tend to have extrapolative expectations over
short-term horizons and mean-reverting or regressive ex-
pectations over longer horizons. If investors have extrapo-
lative expectations over short horizons, they may tend to
accentuate and perpetuate exchange-rate movements in
the short run far beyond the path justified by fundamen-
tals alone. Indeed, if a significant number of investors en-
gaged in extrapolative/trend-following trading strategies,
exchange rates might tend to overshoot on both the up-
side and downside, which could further obscure the rela-
tionship between macroeconomic fundamentals and the
short-term movement of exchange rates.
Because exchange rates can and often do deviate signifi-
cantly from any semblance of fair value in the short run,
investors have looked for alternative forecasting tools to
help them formulate currency investment strategies over
short-term horizons. Short-run forecasting tools that have
attracted interest in recent years include technical-based
trend-following trading rules, sentiment and positioning sur-
veys, FX dealers' customer order flow data, information
embedded in currency option prices, and risk appetite in-
dices.
Investors who concentrate their energies on such tools
presume that the market exhibits a tendency to tip its hand
ahead of time as to which direction it intends to take ex-
change rates in the future. While technical models have
been found to be profitable in the past, most of the other
short-term indicators are relatively new to the FX arena
and thus only a limited time series is available to test their
predictive power. What evidence we do have, however,
suggests that in most cases, these indicators are more
useful as contemporaneous rather than as leading indica-
tors of exchange-rate movements. Nevertheless, they may
prove useful as confirmation indicators that can be used in
conjunction with traditional technical-based trend-follow-
ing trading rules in formulating short-term investment strat-
egies.
Short-Run Forecasting Tools
Short-Run Forecasting Tools: A Checklist
Up
Short-Term Trend
Down
Moving-Average Crossover Trading Rule
Market Sentiment
Speculative Positioning
Order Flow
Option Market Sentiment
Risk Appetite Indices
Neutral Over-
sold
Over-
bought
Momentum
(Consensus Inc. Index of Bullish Opinion)
(Net IMM Contracts)
(DB Customer Order Flow Database)
(Risk Reversals)
Forecasting Tool
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 9
FX market participants typically fall into one of two camps:
(1) Shorter-run technically oriented traders or
(2) Longer-run fundamental-based investors.
Shorter-run technically oriented traders do not base their
investment decisions on fundamental considerations.
Rather, they rely on trend-following trading rules to deter-
mine their position taking: they buy when the currency is
rising, and they sell when the currency is falling. In con-
trast, longer-run fundamental-based investors base their
investment decisions largely on valuation considerations.
If a currency is believed to be mispriced relative to its fair
value, fundamental-based investors would buy those cur-
rencies that are believed to be undervalued and would
sell those currencies that are believed to be overvalued.
Knowing precisely what exchange-rate level represents a
currency's true equilibrium or fair value is not an easy task.
Different exchange-rate models can and often do yield quite
different estimates of a currency's long-run equilibrium
value. In most cases, the marketplace will have only a rough
idea of where a currency's long-run equilibrium value lies.
Because of this, fundamental-based investors will not set
their sights on an imprecise point estimate of fair value,
but rather on an equilibrium range or band. Within this
equilibrium range or band, fundamental-based investors
will presume that the true but unknown equilibrium ex-
change rate, q, lies somewhere between an upper bound,
qU, and a lower bound, qL.
The qU-qL band has been referred to as the "band of agnos-
ticism" in academic writings (see DeGrauwe, 1996). When
exchange rates trade inside the qU-qL band, fundamental-
based investors tend to be agnostic in terms of their cur-
rency positioning, accepting the fact that the actual ex-
change rate is probably trading close to its fair value. Ex-
change-rate movements within the qU-qL band are viewed
as noise and therefore not worthy of serious investment
consideration. Fundamental-based investors would thus
not feel compelled to take on either aggressive long or
short currency positions when exchange rates are trading
inside the band. Instead, they would more likely adopt a
neutral stance toward currency positioning.
When exchange rates fluctuate inside the band of agnosti-
cism, trading tends to be dominated by short-term techni-
cally oriented traders since fundamental-based investors
will refrain from joining the fray until the actual exchange
rate moves outside of the band. When the actual exchange
rate moves outside the qU-qL equilibrium range, fundamen-
tal-based investors will tend to shed their agnosticism and
become more willing to take on aggressive long positions
if the actual exchange rate falls below qL and aggressive
short positions if the actual exchange rate rises above qU.
In times of greater market uncertainty, however, the band
of agnosticism is likely to widen since investor confidence
regarding estimates of fair value is likely to be less strongly
held than in tranquil environments. In such cases, techni-
cally oriented traders would tend to dominate trading ac-
tivity over even wider ranges (q'U-qL'>qU- qL). That, in turn,
would likely lead to even greater FX volatility in the short
run.
One of the key problems for fundamental-based investors
is that even if the exchange rate moved outside of the
band of agnosticism, there is no guarantee that it would
move back inside the band anytime soon. Indeed, funda-
mental-based investors run the risk that an overvalued
currency might get even more overvalued if the exchange
rate moved deeper into overvalued territory, and vice versa.
Since large financial resources are likely to be needed for
investors to position themselves against an overshooting
exchange rate, one has to wonder how many fund manag-
ers would be willing to risk their clients' capital on the
basis of long-run valuation considerations, particularly if
clients evaluate their fund managers investment perfor-
mance over a relatively short time span. If fund managers
view it as simply too risky to take on long or short cur-
rency positions on the basis of long-run valuation consid-
erations, then there might be several occasions when ex-
change rates could wander far from any semblance of fair
value, and yet very few investors would be willing to risk
their clients' capital to bring the exchange rate back into
line with fair value.
Why FX Market Participants Focus Their Energies on Short-Run Rather
Than Long-Run Strategies
The "Band of Agnosticism"
Index of Investor Willingness to
Make Currency Bets Based on
"Fundamentals"
Real Exchange Rate
q’
L
q
L
Analyzing the Behavior of Fundamental-Based Investors When
Expectations of Real Long-Run Equilibrium Exchange Rates Are Loosely Held
q
U
q’
U
Source: Adapted from DeGrauwe (1996)
q
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
10 Deutsche Bank Foreign Exchange Research
When FX dealers were asked recently what role funda-
mental factors played in the determination of exchange
rates, 97% of the respondents felt that fundamentals
played no role on an intra-day basis. However, over me-
dium and longer-term horizons, FX dealers felt that funda-
mental forces did play an important role, with 57.4% of
the respondents believing that exchange rates reflect fun-
damental value on a medium-term (within six months)
basis, and 87% believing exchange rates reflect fundamen-
tal value on a long-term (over six months) basis.
When asked to rank the most important determinants of
exchange rates on an intra-day basis, FX dealers indicated
that bandwagon effects, speculative forces, and over-re-
action to news were the principal driving forces in the very
short run. On a medium-term basis, economic fundamen-
tals and technical trading increase in importance from the
dealer community's perspective, but FX dealers also con-
tinued to assign importance to speculative forces as a key
determinant of medium-term trends in exchange rates. For
longer horizons, FX dealers believed that economic funda-
mentals were the dominant factor driving exchange rates.
However, a not insignificant number (11.3%) believed that
technical trading was important even in the long run.
The FX dealer survey also asked dealers whether they
thought that exchange rates were more predictable on an
intra-day basis or on a medium-term (up to six months) or
long-term (beyond six months) basis. FX dealers were
asked to assign a rating of 1 if there was no predictability,
a rating of 5 if there was a high predictability and a rating
of 2, 3, or 4 if there was low, medium, or better than aver-
age predictability, respectively. Since FX dealers tend to
trade on an intra-day basis, one might have thought that
they would assign a high rating to exchange-rate predict-
ability on an intra-day basis. This was not the case. Indeed,
62% of the dealer respondents gave ratings of 1 or 2, to
the predictability of exchange rates in the short run, while
only 11% gave ratings of 4 or 5.
For medium and longer time horizons, the confidence in
exchange-rate predictability increases, with 30.4% of deal-
ers assigning a ranking of 4 or 5 on a medium-term basis
and 35.1% assigning a ranking of 4 or 5 on a longer-term
basis. The question that we need to ask ourselves is: if FX
dealers are more confident in predicting exchange rates
on a medium/long-term basis rather than on an intra-day
basis, why then do traders concentrate their energies on
very short-run trading? The answer might be that traders
are in a better position to evaluate and manage FX risk on
a short-term basis, which overrides their greater confidence
in medium/long-term exchange-rate predictability.
FX Dealers’ Perception of the Role of Fundamentals
in Explaining Exchange-Rate Movements
FX Dealer Survey Question—Do You Believe Exchange-Rate
Movements Reflect Changes in Fundamental Value on an:
Intraday Medium-Run Long-Run
Basis Basis Basis
(up to 6 months) (beyond 6 months)
Yes 3% 57.8% 87%
No 97% 42.2% 12%
No Opinion 0% 0.0% 1%
Source: Yin-Wong Cheung, Menzie D. Chinn, and Ian W. Marsh,
“How Do UK-Based Foreign Exchange Dealers Think Their Market Operates?”,
NBER Working Paper 7524, February 2000.
FX Dealers’ Perception of the Most Important Factor
That Explains Exchange-Rate Movements
FX Dealer Survey Question—Select the Single Most Important
Factor that Determines Exchange Rate Movements on an:
Intraday Medium-Run Long-Run
Basis Basis Basis
(up to 6 months) (beyond 6 months)
Bandwagon Effects 29.3% 9.5% 1.0%
Over-reaction to News 32.8% 0.7% 0.0%
Speculative Forces 25.3% 30.7% 3.1%
Economic Fundamentals 0.6% 31.4% 82.5%
Technical Trading 10.3% 26.3% 11.3%
Other 1.7% 1.5% 2.1%
Source: Yin-Wong Cheung, Menzie D. Chinn, and Ian W. Marsh,
“How Do UK-Based Foreign Exchange Dealers Think Their Market Operates?”,
NBER Working Paper 7524, February 2000.
FX Dealers’ Perception of the Predictability of
Exchange-Rate Movements
FX Dealer Survey Question—On a Scale of 1 to 5, Indicate If You
Believe the Market Trend Is Predictable on an:
Intraday Medium-Run Long-Run
Basis Basis Basis
(up to 6 months) (beyond 6 months)
1 (Least Predictable) 21.6% 5.9% 17.2%
2 40.3% 20.7% 16.4%
3 26.9% 43.0% 30.6%
4 9.0% 18.5% 20.9%
5 (Most Predictable) 2.2% 11.9% 14.2%
Source: Yin-Wong Cheung, Menzie D. Chinn, and Ian W. Marsh,
“How Do UK-Based Foreign Exchange Dealers Think Their Market Operates?”,
NBER Working Paper 7524, February 2000.
How FX Dealers View the Determination of Exchange Rates in the Short Run
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 11
Technical models generate exchange-rate forecasts by
extrapolating past sequences of exchange-rate movements
into the future. For example, if a currency begins to edge
higher and rises above some critical value, a technical
model will typically issue a recommendation to go long
that currency, the presumption being that the newly formed
trend will continue to carry the currency higher in the fu-
ture. Similarly, a sell signal would be issued if the currency
began to edge lower and fell below some critical value.
Trend-following trading rules will be profitable as long as
the ensuing trend does indeed move in the same direc-
tion as the preceding trend, which would be the case if
exchange rates moved in broad, well-defined cycles. But
that is not to say that exchange rates must always move in
large swings for trend-following trading rules to be profit-
Technical Analysis:
The Advantage of Trading
With and Not Against the Trend
Exchange Rate
Time
Long-Run Equilibrium
Exchange Rate
Technical Analysis:
Even though a currency overshoots its
long-run equilibrium level, a technician
will recommend maintaining long positions
as long as the trend in the exchange
rate is up.
Fundamental Analysis:
May prematurely recommend
selling a currency that has
overshot its long-run
equilibrium level.
Trendline
Market Price
Todays
Spot
Rate
able. What matters for long-term profitability is that large
exchange-rate upswings and downswings must occur on
a frequent enough basis to overcome those occasions
when currency movements are not highly trended.
Fundamental-based models tend to focus on whether a
currency lies above or below its long-run equilibrium or
"fair" value. Technical models, on the other hand, are not
interested in whether a currency lies above or below its
fair value. Rather, technical models are only interested in
whether the trend in the exchange rate is upward or down-
ward. As long as a confirmed uptrend (or downtrend) is in
place, a recommended long (short) position will be main-
tained even if the prevailing trend carries the exchange
rate well above (below) its long-run fair value.
Currency Forecasting Using Technical Analysis—
The Advantage of Trading with and Not Against the Trend
Source: Rosenberg (1996)
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
12 Deutsche Bank Foreign Exchange Research
Identification and Confirmation of a Valid Uptrend/Downtrend
Exchange Rate
Time
Trendline
Market Price
Peak
2
Exchange Rate
Time
Peak
1
Peak
3
Trough
2
Trough
1
Trough
3
Trendline
Peak
2
Peak
3
Trough
2
Trough
1
Trough
3
Peak
1
Market Price
Identification and Confirmation
of a Market Reversal
Exchange Rate
Time
Trendline
Market Price
Head-and-Shoulders Reversal Pattern
Exchange Rate
Time
Neckline
Market Price
Head
Left Shoulder
Right Shoulder
Trend-following trading rules come in many forms. But what
all trend-following trading rules have in common is that
they seek to identify the direction in which the
broad
trend
in exchange rates is heading. One can visualize the trend
in exchange rates as a series of primary and secondary
waves. The primary wave refers to the large, broad moves
in exchange rates that carry the underlying trend in cur-
rency values either upward or downward. The secondary
waves refer to the temporary corrections or partial
retracements of the primary trend that typically take place
over the course of longer-run exchange-rate cycles. What
trend-following trading rules attempt to do is to identify
the direction in which the primary waves are heading.
In a rising market, each rally and partial retracement will
be higher than its predecessor. As long as advancing prices
achieve successively higher peaks and troughs, they indi-
cate that buying pressure is overcoming selling pressure.
In such a case, the uptrend will be presumed to be intact,
until a reversal is signaled. The opposite would be the case
in a declining market.
Markets reverse in many different ways, but all reversals
have one thing in common—all reversals of valid uptrends
(downtrends) must be preceded by the failure of market
prices to achieve successively higher peaks and troughs
(lower troughs and peaks). In the case of a reversal of a
previous valid uptrend, when the wavelike series of rising
peaks and troughs is broken, it indicates that selling pres-
sure is finally beginning to overcome buying pressure.
One of the most recognizable reversal patterns is the head-
and-shoulders pattern. A head-and-shoulders pattern is
essentially a series of three successive rallies with the
second stronger than the first, and the third weaker than
the second. Because the third rally fails to carry as far as
the second, the string of successively higher peaks is bro-
ken. This is an initial indication of weak demand, and the
eventual drop in market prices to levels below the preced-
ing trough (or neckline) is confirmation that a reversal is
taking place. A recent Federal Reserve Bank of New York
study (Osler and Chang, 1995) found that head-and-shoul-
ders patterns have been successful in anticipating rever-
sals in trend for the yen and Deutschemark.
The Identification of FX Market Trends and Reversals
Source: Rosenberg (1996)
Source: Rosenberg (1996) Source: Rosenberg (1996)
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 13
A wide variety of trend-following trading rules abounds,
but by and large they all share the basic property of ex-
trapolation. A chartist who monitors the behavior of mar-
ket prices on a bar chart with a ruler and pencil will as-
sume that an uptrend is intact as long as advancing prices
achieve successively higher peaks and intervening declines
fail to fall below preceding troughs.
More sophisticated extrapolation techniques can be de-
signed with the assistance of a computer. One popular
computer-based technical model is the filter rule, which
issues a buy recommendation if an exchange rate rises by
x%
above its most recent trough, and issues a sell recom-
mendation if the exchange rate falls by
x%
below its most
recent peak. Another popular computer-based technical
model is the moving-average crossover model. By con-
structing longer-run moving averages of daily exchange-
rate movements, one can more easily isolate the primary
trend from short-run noise. If the short-run moving aver-
age of the exchange rate rises above its long-run moving
average, it is an indication that buying pressure is over-
coming selling pressure, and vice versa.
In each case, predictions of the likely future path that ex-
change rates might take are being generated by the ex-
trapolation of the prevailing trend in exchange rates into
the future. Essentially, it really doesn't matter which trend-
following rule you use. Since a trend is a trend, all trend-
following trading rules should roughly generate the same
directional forecast.
Because all of these models generate forecasts by extrapo-
lating the recent past trend of exchange rates into the fu-
ture, buy or sell signals will be issued only after a currency
has already started rising or falling. Although this means
that trend-following trading rules will not capture the very
top and bottom of market moves, they may nevertheless
be profitable if the ensuing exchange-rate movements per-
sist long enough and carry far enough to generate signifi-
cant excess returns.
Filter-Rule Strategy
DM/US$
Time
B
S
P
T
x%
x%
Moving-Average Crossover Strategy
DM/US$
Time
B
S
SRMA
LRMA
All Trend-Following Trading Rules Share
the Same Property: Extrapolation
DM/US$
Time
B
S
Down Trendline
Up Trendline
Trendline Penetration
Trend-Following Trading Rules and the Principle of Extrapolation
Source: Rosenberg (1996)
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
14 Deutsche Bank Foreign Exchange Research
Investors Who Follow a Trend-Following Trading Rule
May Be Vulnerable to Whipsaws Caused by False Signals
DM/US$
Time
B
S
P
T
x%
x%
DM/US$
B
1
S
1
P
1
T
1
An Example of a False Signal
and a Whipsaw Loss
Time
1
Market-Swing
Requirement
Market-Swing Requirement
of Profitable Trend-Following Trading Rules
When exchange-rate swings are sustained and pro-
nounced, the profitability of most technical trading rules
will not be seriously undermined if the trading rule tends
to be a bit late in drawing attention to a shift in market
direction. Unfortunately, this market-swing requirement
could prove in many cases to be a tall order. In markets
that exhibit little overall price variation or are not highly
trended, an investor who adheres to a trend-following trad-
ing rule will find his portfolio vulnerable to potential whip-
saw losses caused by frequent false signals.
Consider the two exchange rate series depicted below.
The series on the left illustrates the working of a profitable
x%
filter rule trading strategy. A recommendation to buy
dollars is issued at point
B
when the DM/US$ exchange
rate rises
x%
from its recent trough at point
T
. The ensuing
market action then carries the DM/US$ exchange rate to a
peak at point
P
, but a recommendation to sell dollars is
issued only at point
S
, when the exchange rate has fallen
x%
from its peak level. An investor who rigidly adhered to
this
x%
filter rule trading strategy would have bought dol-
lars at
B
and sold dollars at
S
, with the spread between
S
and
B
representing the profit margin per dollar invested.
The series on the right illustrates how losses can be in-
curred using a filter rule trading strategy. The series on the
right behaves in a similar fashion as the series on the left,
except for the fact that the dollar's upswing is less pro-
nounced following the recommendation to buy dollars at
point
B
1. That is, the swing in the DM/US$ exchange rate
from
B
1 to
P
1 falls short of the swing from
B
to
P
in the
chart on the left. At
P
1, the dollar's upward momentum is
shown to lose steam far earlier than was the case in the
series on the left. Indeed, the DM/US$ actually begins to
lose ground soon after its initial rise and a sell signal is
eventually issued at point
S
1. Since the selling price (
S
1)
lies below the purchase price (
B
1), a loss is incurred with
the spread between
S
1 and
B
1 representing the margin of
loss per dollar invested. Hence, the buy recommendation
at
B
1 proved to be a false buy signal, with the investor
whipsawed in the process.
Potential Pitfalls from Following Technical Trading Rules
Source: Rosenberg (1996)
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 15
Studies on the long-run profitability of technical-based trad-
ing rules often find that simple moving-average rules would
have generated significant excess returns over relatively
long periods for most major currency pairs. We simulated
more than 2500 sets of moving-average crossover trading
rules for seven currencies versus the U.S. dollar—the Deut-
schemark, Japanese yen, Swiss franc, British pound, and
the dollar-bloc currencies—over the 1986-2002 period, and
unearthed the optimal moving-average trading rules listed
in the table below. The criterion used to choose the opti-
mal trading rule in each case was to identify the rule that
yielded the highest Sharpe ratio (after adjusting for trans-
action costs) for the entire 16-year period.
For each exchange rate, there were a number of moving-
average crossover trading rules that yielded similar attrac-
tive Sharpe ratios, and we list the top 10 trading rules for
each exchange rate. In each case, there was always one
rule that eked out the highest Sharpe ratio and those are
shown in the column on the left. (For example, the best
performing trading rule for the Deutschemark/U.S. dollar
exchange rate was a 1-day/32-day moving-average cross-
over trading rule, which yielded an average annual total
return of 5.01% with a Sharpe ratio of 0.45.) The moving-
average crossover trading rules are ranked for each ex-
change rate and are listed from left to right in the table
below.
The 10 Best Moving-Average Crossover Trading Rules for 1986-2002
(Average Annual Total Returns and Sharpe Ratios for January 1986-April 2002 )
US$ Exchange Rate #1 #2 #3 #4 #5 #6 #7 #8 #9 #10
DEM Mov. Avg. Days 1/32 1/33 1/31 1/28 1/30 1/29 1/34 1/20 1/35 1/19
Total Return (%) 5.01 4.89 4.38 4.20 3.96 3.88 3.98 3.58 3.43 3.13
Std. Dev. (%) 11.06 11.06 11.06 11.06 11.06 11.06 11.06 11.06 11.06 11.06
Sharpe Ratio 0.45 0.44 0.40 0.38 0.36 0.35 0.36 0.32 0.31 0.28
JPY Mov. Avg. Days 8/59 8/60 7/61 8/61 7/60 9/60 9/62 9/61 8/62 8/63
Total Return (%) 9.19 9.16 8.83 8.86 8.76 8.79 8.63 8.63 8.53 8.49
Std. Dev. (%) 11.75 11.75 11.75 11.75 11.75 11.75 11.75 11.75 11.75 11.75
Sharpe Ratio 0.78 0.78 0.75 0.75 0.75 0.75 0.73 0.73 0.73 0.72
GBP Mov. Avg. Days 1/19 1/18 1/20 1/21 1/22 1/16 1/24 1/15 1/17 1/23
Total Return (%) 5.80 5.45 5.36 5.32 5.28 5.20 4.91 4.82 4.80 4.49
Std. Dev. (%) 9.46 9.47 9.46 9.46 9.46 9.47 9.45 9.47 9.47 9.45
Sharpe Ratio 0.61 0.58 0.57 0.56 0.56 0.55 0.52 0.51 0.51 0.48
CHF Mov. Avg. Days 1/57 1/59 1/58 1/56 1/60 1/55 1/69 1/54 1/61 1/65
Total Return (%) 8.63 8.44 8.40 8.40 7.87 7.79 7.50 7.24 7.21 6.84
Std. Dev. (%) 11.65 11.64 11.66 11.65 11.63 11.66 11.63 11.66 11.63 11.64
Sharpe Ratio 0.74 0.72 0.72 0.72 0.68 0.67 0.65 0.62 0.62 0.59
CAD Mov. Avg. Days 14/199 14/200 14/197 13/197 14/198 13/198 15/195 15/193 15/194 14/196
Total Return (%) 1.88 1.88 1.86 1.83 1.82 1.81 1.75 1.66 1.64 1.59
Std. Dev. (%) 4.81 4.81 4.81 4.81 4.81 4.81 4.81 4.81 4.81 4.81
Sharpe Ratio 0.39 0.39 0.39 0.38 0.38 0.38 0.36 0.34 0.34 0.33
AUD Mov. Avg. Days 13/39 13/38 13/40 1/16 13/42 13/41 14/43 13/37 13/43 13/45
Total Return (%) 3.53 3.44 3.41 2.99 2.73 2.54 2.45 2.44 2.37 2.34
Std. Dev. (%) 10.30 10.31 10.28 10.29 10.28 10.28 10.28 10.31 10.28 10.28
Sharpe Ratio 0.34 0.33 0.33 0.29 0.27 0.25 0.24 0.24 0.23 0.23
NZD Mov. Avg. Days 10/17 11/15 10/18 10/15 11/17 14/196 10/16 11/20 11/16 11/14
Total Return (%) 5.20 4.99 5.10 4.76 4.67 4.28 4.28 4.41 3.85 3.38
Std. Dev. (%) 10.79 10.80 10.78 10.80 10.79 10.14 10.80 10.78 10.80 10.81
Sharpe Ratio 0.48 0.46 0.47 0.44 0.43 0.42 0.40 0.41 0.36 0.31
Note: A Sharpe Ratio measures the amount of return on an investment (less the return of a risk-free asset) per unit of risk, which is proxied by its standard deviation.
Datastream is the source of the underlying exchange-rate data.
Empirical Evidence on the Profitability of Moving-Average Trading Rules
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
16 Deutsche Bank Foreign Exchange Research
Given the Sharpe ratios shown in the table on the preced-
ing page, one might wonder why more investors do not
rely on trend-following trading rules more regularly. One
reason is that investors might have a fairly high Sharpe
ratio threshold for committing capital to any particular trad-
ing strategy. For instance, many investors might not em-
brace a trading strategy rule unless it generated a Sharpe
ratio of at least 0.50. And in some cases, an even higher
Sharpe ratio might be required for an investor to add sig-
nificant risk capital to a particular trading rule. If that were
the case, only the yen and Swiss franc would qualify as
currencies worthy of trading from a technical perspective.
While the estimated Sharpe ratios may have been high on
average for the 1986-2002 period, one should note that
the Sharpe ratios were highly volatile when viewed on a
one-year rolling basis. As the charts below show, there
were periods when risk-adjusted return performances were
high and other periods when risk-adjusted return perfor-
mances were downright poor. This was clearly the case
for several key exchange rates in 2001, with one-year roll-
ing Sharpe ratios falling into negative territory for the Deut-
schemark, British pound, and Swiss franc.
Except for the yen and the dollar-bloc currencies, total re-
turn performances were not that impressive in 2001. In
most cases, currencies were largely range-bound versus
the U.S. dollar in 2001 and therefore generated a consider-
able number of false signals that yielded small but fre-
quent losses. On the other hand, the yens downtrend—
particularly in late 2001—was clearly evident and exploit-
able by following a moving-average crossover trading rule.
Deutschemark Moving-Average Crossover
Trading Rule Risk-Adjusted Performance
(1993-2002)
-3
-2
-1
0
1
2
3
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
(One-Year Rolling Sharpe Ratio)
Japanese Yen Moving-Average Crossover
Trading Rule Risk-Adjusted Performance
(1993-2002)
-2
-1
0
1
2
3
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
(One-Year Rolling Sharpe Ratio)
British Pound Moving-Average Crossover
Trading Rule Risk-Adjusted Performance
(1993-2002)
-3
-2
-1
0
1
2
3
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
(One-Year Rolling Sharpe Ratio)
Swiss Franc Moving-Average Crossover
Trading Rule Risk-Adjusted Performance
(1993-2002)
-2
-1
0
1
2
3
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
(One-Year Rolling Sharpe Ratio)
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 17
Most academic studies as well as our own work docu-
ment the fact that moving-average trading rules have
tended to generate attractive risk-adjusted returns over the
long run. What is not often pointed out in these studies is
the frequency with which winning and losing positions tend
to occur. As the table below illustrates, moving-average
trading rules have tended to generate far more losing than
winning trades on average. For example, in the case of
the Deutschemark, Swiss franc, cable, and the A$, roughly
75% of the recommended trades generated by optimized
moving-average trading rules end up as losing trades. Only
25% of the trades proved to be winning ones. In the case
of the C$ and NZ$, roughly two-thirds of the recommended
trades resulted in losses. Only in the case of the yen was
the win/loss ratio close to 50%.
From a longer-run perspective, having a trading rule that
generates such frequent losing trades does not pose a
serious problem if the losing positions are cut quickly—so
only a small loss is taken—and profits are allowed to ride
on the less frequent correctly predicted winning trades.
Indeed, this is often what is found to be the case. As shown
in the table below, the average profit on the less frequent
winning trades has tended to exceed the average loss on
the more frequent losing trades by a fairly hefty margin.
This margin of difference has been sufficient to allow mov-
ing-average trading rules to be profitable in the long run.
Frequency of Winning and Losing Trades Generated by Moving-Average
Trading Rules
Profit and Losses from Optimal Moving-Average Trading Rules
(Risk-Return Analysis & Winning versus Losing Trades of Selected Currencies versus the U.S. Dollar)
(January 1986-April 2002)
DEM JPY GBP CAD AUD NZD CHF
Optimal Moving-Average Trading Rule
Number of Days (SRMA/LRMA) 1/32 8/59 1/19 14/199 1/16 10/17 1/57
Average Annual Return 5.0% 9.2% 5.8% 1.9% 3.5% 5.2% 8.6%
Standard Deviation of Returns 11.1% 11.8% 9.5% 4.8% 10.3% 10.8% 11.7%
Sharpe Ratio 0.45 0.78 0.61 0.40 0.34 0.48 0.74
Total Recommended Trades 342 81 452 35 218 165 217
Winning Trades 91 37 126 13 61 58 57
Winning Trade Percentage 27% 46% 28% 37% 28% 35% 26%
Losing Trades 251 44 326 22 157 107 160
Losing Trade Percentage 73% 54% 72% 63% 72% 65% 74%
Average Profit on Winning Trades 2.97% 5.92% 2.27% 3.69% 2.03% 2.11% 4.63%
Average Profit on Losing Trades -0.77% -1.72% -0.62% -1.03% -0.63% -0.84% -0.84%
Ratio of Profits/Losses 3.86 3.44 3.66 3.58 3.22 2.51 5.51
Note: A Sharpe Ratio measures the amount of return on an investment (less the return of a risk-free asset) per unit of risk,
which is proxied by its standard deviation.
Datastream is the source of the underlying exchange-rate data.
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
18 Deutsche Bank Foreign Exchange Research
The charts below reveal the distribution of returns that each
of our optimal moving-average crossover trading rules has
generated over the past 16 years. As shown, the over-
whelming majority of recommended trades resulted in ei-
ther small gains or small losses. However, note that the
distribution of returns is skewed heavily to the right in most
cases. This indicates that large positive returns do occur
from time to time; it is just that the frequency of big win-
ning trades tends to be quite low. These highly skewed
return distributions raise the question whether conventional
measures of risk such as standard deviation, Sharpe ra-
tios, and information ratios accurately convey the asym-
metric risks facing technically oriented investors, particu-
larly over short horizons.
Distribution of Returns from Moving-Average Trading Rules
Distribution of Returns of DEM Trades
(Generated by Moving-Average Crossover Strategy)
(1986-2002)
0
50
100
150
200
-2.9
-2.0
-1.1
-0.20.71.6 2.5 3.4 4.35.26.1 7.0 7.9 8.89.7
10.6
11.5
12.4
13.1
Frequency
Return(%)
Distribution of Returns of JPY Trades
(Generated by Moving-Average Crossover Strategy)
(1986-2002)
0
10
20
30
40
50
-5.4 -2.1 1.3 4.6 8.0 11.4 14.7 18.1 21.4 24.5
Frequency
Return(%)
Distribution of Returns of GBP Trades
(Generated by Moving-Average Crossover Strategy)
(1986-2002)
0
50
100
150
200
250
300
-4.1
-2.8
-1.4
-0.1
1.2
2.5
3.8
5.2
6.5
7.8
9.1
10.4
11.8
13.1
14.4
15.7
17.0
18.3
19.7
21.0
22.3
23.6
Frequency
Return(%)
Distribution of Returns of CAD Trades
(Generated by Moving-Average Crossover Strategy)
(1986-2002)
0
5
10
15
20
25
30
-2.46 0.81 4.07 7.34 10.6 13.6
Frequency
Return(%)
Distribution of Returns of AUD Trades
(Generated by Moving-Average Crossover Strategy)
(1986-2002)
0
20
40
60
80
-2.4 -1.7 -1.0 -0.4 0.3 1.0 1.7 2.3 3.0 3.7 4.4 5.1 5.7 6.4 6.8
Frequency
Return(%)
Distribution of Returns of CHF Trades
(Generated by Moving-Average Crossover Strategy)
(1986-2002)
0
20
40
60
80
100
120
140
-5.73
-4.13
-2.52
-0.91 0.69 2.3 3.9 5.51 7.12 8.72
10.33
11.93
13.54
15.15 16.2
Frequency
Return(%)
Source: Datastream
Source: Datastream
Source: Datastream
Source: Datastream
Source: Datastream
Source: Datastream
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 19
The charts below provide a different way of looking at the
short-term risks facing technically oriented investors in the
FX arena. These charts show the actual gains and losses
on the recommended trades that our optimal moving-av-
erage trading rules have generated in recent years. What
these charts reveal is that there were often frequent runs
of successive losing trades that had to be absorbed be-
fore a sizeable winning trade was generated. While posi-
tive returns were generated in most cases for the entire
sample period, clearly there were uncomfortably lengthy
intervals when frequent losses were incurred.
Our analysis raises several interesting issues. First, should
technical-based moving-average trading rules be viewed
as short-term forecasting tools or instead would it be bet-
ter to view such trading rules as long-term forecasting
tools? The results here suggest that technical-based mov-
ing-average trading rules work best in the long run, not in
the short run. Indeed, the odds that a moving-average trad-
ing rule recommendation will generate a profit appear to
be no better than 25%-35% in most cases in the short
run. Second, it is possible that since the frequency of los-
ing trades is so large, this might actually dissuade inves-
tors from using technical models in formulating currency
investment strategies. If investors shy away from using
technical models because of the high frequency of short-
term losses, this might explain why excess returns in the
long run from technical-based models are not completely
arbitraged away.
In order to successfully trade currencies using a technical-
based moving-average trading rule, an investor needs to
have staying power—considerable risk capital on hand to
absorb possible frequent short-term trading losses—and
patience. There might be other technical tools that could
be used to cut down on the number of frequent losses on
short-term trading positions, but one needs to be mindful
of the fact that attempts to add "filters" or other types of
technical bells and whistles to limit the frequency of false
signals might hinder the upside potential of correctly pre-
dicted winning trades.
Should Moving-Average Trading Rules Be Viewed as Short or Long-Run
Forecasting Tools?
Profit and Losses on Individual DEM Trades
(Generated by Moving-Average Crossover Strategy)
(Most Recent 140 Trades)
-4
-2
0
2
4
6
8
10
12-27-95 04-21-97 03-27-98 01-14-99 11-08-2000 10-02-2001
Returns(%)
Trade Dates
Profit and Losses on Individual JPY Trades
(Generated by Moving-Average Crossover Strategy)
(1986-2002)
-10
-5
0
5
10
15
20
25
30
01-01-87 12-27-89 02-02-94 01-23-98 03-07-2002
Returns(%)
Trade Dates
Profit and Losses on Individual GBP Trades
(Generated by Moving-Average Crossover Strategy)
(Most Recent 140 Trades)
-2
-1
0
1
2
3
4
5
6
06-12-97 06-01-98 01-27-99 01-03-2000 01-12-2001 09-24-2001
Returns(%)
Trade Dates
Profit and Losses on Individual CAD Trades
(Generaged by Moving-Average Crossover Strategy)
(1986-2002)
-5
0
5
10
15
20
11-03-86 06-23-89 11-02-90 09-16-94 04-25-96 01-14-97 08-06-99
Returns(%)
Trade Dates
Datastream is the source of the
underlying exchange-rate data.
Datastream is the source of the
underlying exchange-rate data.
Datastream is the source of the
underlying exchange-rate data.
Datastream is the source of the
underlying exchange-rate data.
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
20 Deutsche Bank Foreign Exchange Research
Most academic studies have concluded that exchange-rate
movements closely approximate a random walk process.
A random walk (serial correlation) test seeks to investi-
gate whether there exists a positive, negative, or zero lin-
ear relationship between today’s change in the exchange
rate and yesterday’s change in the exchange rate. Evidence
of a positive linear relationship would indicate the exist-
ence of trend persistence since a positive change in a
currency's value yesterday would tend to be followed by a
positive change today.
Estimated serial correlation coefficients could vary between
+1, 0, and -1 depending on whether there exists a strong
positive relationship, no relationship, or a strong negative
relationship between successive exchange-rate changes.
The weight of evidence generally supports the view that
for most currencies, the estimated serial correlation coef-
ficients are often quite small and in many cases statisti-
cally insignificant from zero.
Although researchers often find support for the view that
exchange-rate movements tend to fluctuate randomly on
a daily basis, they also find evidence that exchange-rate
changes are positively serially correlated when viewed on
a monthly basis. Thus, although exchange rates may fluc-
tuate randomly over very short time spans (i.e., daily), they
tend to rise and fall on a trend basis on a medium/long-
term (i.e., monthly) basis. If so, this would imply that trend-
following trading rules could be devised to profit from these
medium/long-term trends that exchange rates follow.
One of the problems with serial correlation tests is that
they seek only to determine whether a stable "linear" rela-
tionship exists between successive exchange-rate move-
ments. Although it might be the case that successive ex-
change-rate changes are linearly independent, they might
nevertheless exhibit significant positive nonlinear depen-
dence, which traditional serial correlation tests would not
detect. This might explain why researchers have found that
exchange rates follow a random walk, yet at the same time
have found that trend-following technical trading rules have
been profitable. It may be the case that there does exist
some form of serial dependency in successive exchange-
rate movements; it's just that this serial dependence is
not linear. If there exists some form of nonlinear depen-
dence in successive exchange-rate movements, it would
appear that trend-following trading rules are capturing this
form of serial dependence.
Random Walk Tests and the Profitability of Technical Trading Rules
(a)
Positive Serial Correlation
Serial Correlation Tests:
Examining Whether a Stable Linear Relationship Exists between
Current and Previous Changes in an Exchange Rate
(b)
Zero Serial Correlation (c)
Negative Serial Correlation
E
t
E
t
E
t
E
t-1
E
t-1
E
t-1
Source: Adapted from Fogler (1978)
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 21
Fund managers have turned to sentiment and positioning
surveys in recent years as forecasting tools. In most cases,
exchange rates move in the same direction that sentiment
and positioning surveys are moving in the short run. For
example, when sentiment toward the dollar becomes in-
creasingly bullish—as reported in the weekly Consensus
Inc. Survey of Bullish Market Opinion—the dollar tends to
rise. Similarly, when long U.S. dollar speculative positions
are undertaken on the IMM (money-market futures ex-
change), the dollar tends to rise as well.
What does the empirical evidence say about the success
of sentiment/positioning surveys in predicting exchange-
rate movements? Our analysis, listed in the table below,
finds that changes in sentiment and positioning surveys
are statistically significant, for the most part, in explaining
contemporaneous changes in exchange rates, but that
lagged values of sentiment and positioning survey data
are statistically insignificant in predicting current changes
in exchange rates. This suggests that sentiment/position-
ing survey data cannot be relied upon as forecasting tools.
What does the empirical evidence say about the useful-
ness of sentiment/positioning survey data as contrarian
indicators? The trends in market sentiment and specula-
tive positioning could be viewed as potential overbought
or oversold indicators if the rise or fall in each of those
series appeared stretched relative to historical norms.
Unfortunately, the empirical evidence suggests that there
is no statistically significant relationship between over-
stretched sentiment and positioning surveys and subse-
quent changes in exchange rates. Based on these find-
ings, it could prove risky to adopt a contrarian position sim-
ply because investors' FX exposures appeared to be at
extreme levels. History is replete with examples where
over-stretched markets have stayed over-stretched for a
considerable time. Perhaps all that one can say is that if
one or more of those indicators moved into significantly
overbought or oversold territory, then the
odds
of a rever-
sal would have increased, but that it would not be pos-
sible to issue a definitive signal that an imminent reversal
was at hand.
Although trends in sentiment and positioning surveys may
not help us in forecasting future movements in exchange
rates, they may nevertheless be useful as trend confirma-
tion indicators. Because of the strong positive contempo-
raneous correlation between the sentiment and position-
ing data, on the one hand, and the trend in exchange rates
on the other, sentiment and positioning data could be used
in conjunction with a moving-average model to confirm
whether an exchange-rate uptrend or downtrend is intact.
Market Sentiment and the U.S. Dollar
(Consensus Inc. Index of Bullish Opinion)
(1998-1999)
90
92
94
96
98
100
102
30
40
50
60
70
80
90
Jan-98 Apr-98 Jul-98 Oct-98 Jan-99 Apr-99 Jul-99 Oct-99
US$ Index----
A
nalysts Bullish on the US$(4-Week Avg.)(%)____
Sources: Datastream;
Based on Consenus Inc. Index of Bullish Market Opinion
Market Sentiment and the U.S. Dollar
(Consensus Inc. Index of Bullish Opinion)
(April 2000-November 2001)
94
96
98
100
102
104
106
108
30
40
50
60
70
80
90
Apr-00 Jul-00 Oct-00 Jan-01 Apr-01 Jul-01 Oct-01
US$ Index----
A
nalysts Bullish on the US$(4-Week Avg.)(%)____
Based on Consenus Inc. Index of Bullish Market Opinion
By permission of Consensus, Inc., (1) (816) 373-3700,
Consensus, National Futures and Financial Weekly,
www.consensus-inc.com
Speculative Positioning and the U.S. Dollar
(IMM Derived Net Non-Commercial FX Positions)
(April 2000-November 2001)
94
96
98
100
102
104
106
108
-150
-100
-50
0
50
100
Apr-00 Jul-00 Oct-00 Jan-01 Apr-01 Jul-01 Oct-01
US$ Index----
S
um of Net Contracts vs. USD(000s)____
Assessing the Statistical Significance of
Sentiment/Positioning/Flow Variables as
Explanatory Variables of Exchange-Rate Movements
(January 15, 1999-November 2, 2001 Weekly Data)
Lagged
Contemporaneous One Period
Consensus Inc. Bullish Market Opinion (US$)
Regression Coefficient 0.12 0.09
(t-Statistic) (3.90) * (1.15)
IMM Net Non-Commercial FX Positions (US$)
Regression Coefficient 0.23 0.01
(t-Statistic) (5.65) * (0.24)
*Indicates a statistically significant value at the 95% confidence level.
By permission of Consensus Inc.
Are Sentiment and Positioning Indicators Useful in Predicting FX Movements?
Source: IMM, Datastream
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
22 Deutsche Bank Foreign Exchange Research
Investors often try to infer the market's expectation of fu-
ture interest-rate movements by examining interest-rate
futures contracts for various maturities or by examining
the implied future path in interest rates embedded in yield-
curve slopes. Similarly, investors often try to infer the
market's expectation of future exchange-rate movements
by examining the level of domestic/foreign interest-rate
spreads, or the implied path of long-dated forward ex-
change rates.
Information on future interest-rate and exchange-rate move-
ments can also be gleaned from the options market.
Whereas interest-rate futures, yield-curve slopes, and yield
spreads provide point estimates regarding expected fu-
ture values, the options market provides a different piece
of information—the market's expectation of the probabil-
ity distribution of future interest rates and exchange rates.
For example, in the FX market, data on implied volatility
for a range of maturities and strike prices are readily avail-
able. Implied volatility provides us with a measure of the
marketplace's uncertainty regarding future exchange-rate
movements. If implied volatility is rising, it "implies" greater
uncertainty about future exchange-rate movements, and
vice versa.
One could construct a forward implied volatility curve to
glean expected future volatility movements by stringing
together implied volatilities for a range of maturities with
the same strike price. If the forward implied volatility curve
were steeply upward sloping, the market would be pricing
in the expectation of a future jump in currency volatility. If
the forward implied volatility curve were flat, the market-
place would be pricing in no change in the level of cur-
rency volatility in the future.
Does the forward implied volatility curve accurately antici-
pate future jumps in exchange-rate volatility? Because vola-
tility jumps often occur suddenly and without warning, it
is highly unlikely that the forward implied volatility curve
would be able to consistently forecast future jumps in the
exchange rate. Indeed, the Bank of England examined
whether implied forward volatility curves anticipated the
unwinding of the infamous yen carry (long dollar/short yen)
trade in October 1998, and concluded that "Forward vola-
tility curves…. did not expect the increase in volatilities
(that occurred) in October 1998. Although (historical) vola-
tility had increased throughout the summer, the forward
volatility curve suggested that it would fall back towards
previous levels." Indeed, in mid-September 1998, histori-
cal volatility was running around 22%-26% and forward
volatility curves were expecting implied volatilities to ease
toward 14% in the October-November 1998 period. Instead,
historical volatility soared to over 40% in early October
when the yen carry trade was suddenly unwound.
Can One Extract Information from the Currency Options Market
to Predict Future Exchange-Rate Movements?
Japanese Yen/U.S. Dollar Exchange Rate
(1997-1998)
110
120
130
140
150
Jan-97 Apr-97 Jul-97 Oct-97 Jan-98 Apr-98 Jul-98 Oct-98
Yen/US$
Japanese Yen/U.S. Dollar Implied Volatility
(1997-1998)
0
10
20
30
40
Jan-97 Apr-97 Jul-97 Oct-97 Jan-98 Apr-98 Jul-98 Oct-98
1-Month Volatility----(%) 12-Month Volatility____
Source: Datastream
Source: DRI
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 23
Assessing the Statistical Significance of
Sentiment/Positioning/Flow Variables as
Explanatory Variables of Exchange-Rate Movements
(January 15, 1999-November 2, 2001 Weekly Data)
Lagged
Contemporaneous One Period
Option Market Sentiment (Euro Risk Reversals)
Regression Coefficient -0.02 0.03
(t-Statistic) (-0.90) (1.23)
Option Market Sentiment (Yen Risk Reversals)
Regression Coefficient 0.11 0.04
(t-Statistic) (3.06) * (1.19)
*Indicates a statistically significant value at the 95% confidence level.
Source: Datastream; DB
Euro Risk Reversals & US$/Euro Exchange Rate
Euro Calls(+)/Puts(-) Trading at Premium vs. USD
0.82
0.84
0.86
0.88
0.90
0.92
0.94
0.96
0.98
-1.0
-0.5
0.0
0.5
1.0
1.5
Jul-00 Oct-00 Jan-01 Apr-01 Jul-01 Oct-01 Jan-02 Apr-02
US$/Euro----
C
alls(+)/Puts(-) Trading at Premium____
Yen Risk Reversals & Yen/US$ Exchange Rate
Yen Calls(+)/Puts(-) Trading at Premium vs. USD
(140)
(135)
(130)
(125)
(120)
(115)
(110)
(105)
(100)
-2.0
-1.0
0.0
1.0
2.0
3.0
Jul-00 Oct-00 Jan-01 Apr-01 Jul-01 Oct-01 Jan-02 Apr-02
Yen/US$(reverse scale)----
C
alls(+)/Puts(-) Trading at Premium____
FX traders often use risk reversals to glean information on
whether the market might be attaching a higher probabil-
ity to a large currency appreciation than to a large currency
depreciation, or vice versa. A risk reversal is a currency
option position that consists of the purchase of an out-of-
the-money (25
delta
) call and the simultaneous sale of an
out-of-the-money (25
delta
) put, both in equal amounts and
both with the same expiration date. Risk reversals are
quoted in terms of the implied volatility spread between
the 25
delta
call and 25
delta
put. For example, if the im-
plied volatility on the call were 2% larger than the implied
volatility on the put, the risk reversal would be quoted at
+2.0%. If the implied volatility on the put were 2% greater
than the implied volatility on the call, the risk reversal would
be quoted at -2%.
A risk reversal quoted at +2% would indicate that the
market was attaching a higher probability to a large cur-
rency appreciation than to a large currency depreciation.
This would indicate that the market was willing to pay more
to insure against the risk that the currency will rise sharply
than it was willing to pay to insure against the risk that the
currency will fall sharply.
The key issue for traders and investors is whether the level
or trend in currency risk reversals can be used to correctly
anticipate future exchange-rate movements. The evidence
indicates a high contemporaneous correlation between the
trend in risk reversals and the trend in exchange rates, but
no statistically significant relationship exists between
lagged risk reversal data and future exchange-rate move-
ments. Therefore, risk reversals are capable of confirming
an exchange rate’s trend, but not predicting it.
Nor is there evidence that overly stretched risk reversal
measures can function as a consistently reliable contrary
indicator. Indeed, the Bank of England's study on the un-
winding of the yen carry trade in the fall of 1998 found that
dollar/yen risk reversals failed to provide an early warning
of the dramatic unwinding of long dollar/short yen posi-
tions that was about to occur.
Inferring Information about Future Exchange-Rate Movements
from Currency Risk Reversals
Yen/US$
Profit
Payoff Schedule of 25 Delta
Risk Reversal at Maturity
(-)
(+)
0
Source: Datastream; DB Source: Datastream; DB
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
24 Deutsche Bank Foreign Exchange Research
In recent years, there has been increased interest on the
part of investors and academicians in FX-dealer customer-
flow data. One of the characteristics that distinguishes the
FX market from the world equity market is that the FX
market has considerably less transparency. Equity-market
disclosure requirements mandate that all trades are in-
stantly posted. Thus, volume and price data are instantly
available to all parties. Not so in the FX market. No such
disclosure requirements exist, which means that order-flow
information is not immediately available to all parties.
Indeed, large FX dealers are in a unique position in that
they are able to observe large trades that could move the
market before other parties are aware of such trades. A
recent survey of FX dealers indicated that most traders
believe that large FX dealers have a comparative advan-
tage over smaller dealers because they have a large cus-
tomer base that gives them privileged information about
customer orders.
One could envisage a marketplace where there exists both
private and publicly available information. Publicly available
information consists largely of macroeconomic-related
data, which is available to all market participants at the
same time. However, there may be microeconomic-related
information that is important for exchange rates, but is not
publicly available. Examples of microeconomic-related in-
formation are shifts in risk appetite, liquidity needs, hedg-
ing demands, and institutional portfolio rebalancings. Such
microeconomic-related FX flows give rise to buy or sell
orders that influence the short-term trend in exchange
rates, just as investor responses to macroeconomic-related
information do. Large FX dealers can use the information
gleaned from the change in their customer order flow to
drive their own short-run strategies, which can then add to
the upward or downward pressure on the exchange rate's
value.
Viewing the impact of order flow in this manner, it is evi-
dent that order flow serves as a proximate determinant
and not as the ultimate determinant of short-term ex-
change-rate movements. The ultimate determinant of ex-
change rates is the joint interplay of macroeconomic and
microeconomic information underlying the change in or-
der flow.
Given the potential impact that order flow can have on
exchange rates in the short run, a large FX dealer's order
book can provide valuable information to both traders and
investors who wish to keep abreast of underlying private
investor shifts in portfolio behavior, whether those shifts
are of a macro or micro-economic nature.
Information
Not Publicly
Available
Portfolio
Shifts Order
Flow
Change
in
Exchange Rate
The Effect of Information on Order Flow
and Exchange Rates
Publicly
Available
Information
Reasons for Competitive Advantage
of Large FX Players
(Based on an FX Dealer Survey)
Survey Response (%)
Large Customer Base 33.3
Better Information 22.9
Deal in Large Volumes 14.8
Ability to Affect Exchange Rate 9.4
Offer New FX Products 6.2
Access to Global Trading Network 4.7
Experienced Trades 4.2
Lower Costs 2.9
Smaller Counterparty Risk 0.5
Other 1.0
Source: Cheung and Chinn, “Traders, Market Microstructures and
Exchange Rate Dynamics”, unpublished paper.
FX Dealer Order Flow and the Determination of Exchange Rates
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 25
How FX Dealers and Fund Managers View the Importance of
Flow Analysis Relative to Fundamental and Technical Analysis
According to a recent survey of professional FX dealers
and fund managers, both FX dealers and fund managers
treat the analysis of foreign-exchange flows as a distinct
form of analysis, independent from more traditional fun-
damental and technical approaches. A recent survey by
Gehrig and Menkhoff (2002) indicates that FX market par-
ticipants in general attach the greatest weight to technical
analysis (40.2%), followed by fundamental analysis
(36.3%), and finally by flow analysis (23.5%). Among the
participants surveyed, FX dealers were the ones who as-
signed the most weight to flow analysis, with 26.2% of
them viewing it as their most important source of informa-
tion. By comparison, only 16.8% of fund managers re-
garded flow analysis as their most important source of
information, with 46.2% considering fundamental analy-
sis as their most important source.
FX market participants generally regard the analysis of FX
flows as more useful for short-term rather than long-term
forecasting. As shown in the pie chart on the bottom right,
25.4% of all respondents believe that flow analysis pro-
vides valuable information only on an intra-day basis. An-
other 37.3% believe that the information provided by flow
analysis can be useful up to a few days. Combined, 62.7%
of all respondents believe that flow analysis should be lim-
ited to forecasting horizons up to only a few days. In terms
of the longer-run usefulness of flow analysis for currency
forecasting, 21.9% of FX respondents believe that flow
analysis can be useful for several weeks, while only a
modest 15.4% believe that flow analysis can be useful for
up to several months into the future.
Survey of All FX Market Participants
The Importance of Fundamental, Technical & Flow
Analysis in Currency Investment Decision
Fundamental Analysis--36.3%
Technical Analysis--40.2%
Flow Analysis--23.5%
Survey of Fund Managers
The Importance of Fundamental, Technical and
Flow Analysis in Currency Investment Decision
Fundamental Analysis--46.2%
Technical Analysis--37.0%
Flow Analysis--16.8%
Survey of FX Dealers
The Importance of Fundamental, Technical & Flow
Analysis in Currency Investment Decision
Fundamental Analysis--32.4%
Technical Analysis--41.4%
Flow Analysis--26.2%
The Usefulness of Flow Analysis
at Different Forecasting Horizons
(Survey of All FX Participants)
Intraday--25.4%
A Few Days--37.3%
A Few Weeks--21.9%
More than 2 Months--15.4%
Source: Gehrig and Menkhoff (2002) Source: Gehrig and Menkhoff (2002)
Source: Gehrig and Menkhoff (2002) Source: Gehrig and Menkhoff (2002)
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
26 Deutsche Bank Foreign Exchange Research
Like positioning and sentiment data, the empirical evidence
suggests that FX customer order flow has a contempora-
neous—rather than a predictive—relationship with ex-
change-rate movements. The chart below indicates that
cumulative trends in order flow and the trend in the dollar
closely parallel one another on a contemporaneous basis.
(Note that since both exchange rates and customer flow
data can fluctuate erratically on a daily basis, we have
smoothed out the erratic daily fluctuations in these time
series to discern whether a strong positive correlation ex-
ists between order flow and exchange rates.)
There is no evidence, however, of a statistically significant
relationship between lagged order flow and future ex-
change-rate movements. This finding is supported by vari-
ous academic studies that have investigated the impor-
tance of order flow on exchange rates.
Such findings should not come as a surprise. After all, if
there are more dollar buyers than dollar sellers on a sus-
tained basis in the FX market, both cumulative net dollar
purchases and the dollar's value should rise in a parallel
fashion. One should expect dollar purchases in the cur-
rent period to drive the dollar higher now, not at some
point in the future.
Empirical work on customer order flow's impact on ex-
change rates indicates that certain customers' FX orders
may exert a greater impact on exchange rates than other
customers' FX orders. Research by Fan and Lyons (1999)
indicates that real-money accounts (i.e., unleveraged fund
managers) played a more important role in driving the euro
than did corporate or hedge fund players. In the case of
the yen, hedge funds and corporates played a relatively
more important role.
Assessing the Statistical Significance of
Sentiment/Positioning/Flow Variables as
Explanatory Variables of Exchange-Rate Movements
(January 15, 1999-November 2, 2001 Weekly Data)
Lagged
Contemporaneous One Period
Market Positioning in the US$ (DB Flow Data)
Regression Coefficient 0.69 -0.40
(t-Statistic) (1.97) * (-1.13)
Market Positioning in the Euro (DB Flow Data)
Regression Coefficient 0.066 0.057
(t-Statistic) (2.88) * (1.82)
Market Positioning in the Yen (DB Flow Data)
Regression Coefficient 0.187 -0.024
(t-Statistic) (1.75) (-0.22)
*Indicates a statistically significant value at the 95% confidence level.
The Impact of Real Money, Hedge Fund, and Corpo-
rate Customer Order Flows on Exchange Rates
(January 1993-June 1999, Monthly Data)
Real Money Leveraged Non-Financial
(Unhedged) (Hedged) Corporate
Flow Flow Flow
Euro (R2 = 0.27)
Regression Coefficient 1.5 0.6 -0.2
(t-Statistic) (4.6) (1.6) (-0.5)
Yen (R2 = 0.34)
Regression Coefficient 1.1 1.8 -2.3
(t-Statistic) (1.9) (4.9) (-3.5)
*Source: Richard K. Lyons,
The Microstructure Approach to Exchange Rates
, 2001.
Market Positioning and the U.S. Dollar
(DB Flow Momentum Indicator, Moving Average)
(April-November 2001)
102
103
104
105
106
107
108
-10
-5
0
5
10
Apr-01 May-01 Jun-01 Jul-01 Aug-01 Sep-01 Oct-01 Nov-01
US$ Index----Net USD Buying/Selling(index)___
Source: DB, Datastream
Customer Order Flow and Exchange Rates—Empirical Evidence
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 27
Surveys of global fund-manager positioning are another
tool that FX market participants have turned to in recent
years for short-term exchange-rate forecasting. The monthly
Russell-Mellon survey summarizes the portfolio positions
of 28 global fixed-income managers in selected curren-
cies and bond markets, with attention normally focused
on the 25th, median, and 75th percentile exposures. The
trends in investor exposure to the dollar, euro, and yen for
the 1996-2001 period are shown in the charts on the left.
In an attempt to better detect underlying trends in currency
positioning, we have calculated a net weighted-average
exposure index for investor allocations to the dollar, euro,
and yen in the Russell-Mellon survey. We make the heroic
assumption that the range of all reported currency expo-
sures can be defined by a normal distribution, and that the
25th, median, and 75th percentile exposure fit comfortably
inside this normal distribution. We then calculate a net
weighted-average exposure for each currency, which is
plotted against each of the respective currencies in the
diagrams on the right. These charts suggest that the use-
fulness of fund-manager positioning in predicting exchange-
rate movements may be tenuous at best.
Investor Positioning and the Trend in Exchange Rates
Source: Russell Mellon, Datastream
Source: Russell Mellon, Datastream
Source: Russell Mellon, Datastream
Investor Exposure to the U.S. Dollar
(DB Net Weighted-Average Exposure to the US$)
(Russell-Mellon Survey)
-20
-10
0
10
20
30
40
1996 1997 1998 1999 2000 2001 2002
25 and 75 Percentiles----
Net Weighted Exposure Index____
Source: Russell Mellon
Investor Exposure to the Euro
(DB Net Weighted-Average Exposure to the Euro)
(Russell-Mellon Survey)
-30
-20
-10
0
10
20
1996 1997 1998 1999 2000 2001 2002
25 and 75 Percentiles----
Net Weighted Exposure Index____
Source: Russell Mellon
Investor Exposure to the Japanese Yen
(DB Net Weighted-Average Exposure to the Yen)
(Russell-Mellon Survey)
-20
-15
-10
-5
0
5
1996 1997 1998 1999 2000 2001 2002
25 and 75 Percentiles----
Net Weighted Exposure Index____
Source: Russell Mellon
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
28 Deutsche Bank Foreign Exchange Research
Surveys of investor positioning can also be used to gauge
investors' appetite to take on currency risk. We have con-
structed a composite currency exposure index using the
net weighted-average exposures of the Russell Mellon sur-
vey. The Deutsche Bank Currency Exposure Index is a
simple average of the absolute value of the net weighted-
average overweight/underweight exposures that global
fund managers maintain toward the dollar, euro, and yen.
A high index value indicates that overall investor position-
ing is aggressive, while a low index value indicates that
overall investor positioning is not aggressive.
The Deutsche Bank Currency Exposure Index reveals that
investors willingly took on aggressive currency exposures
in 1996-97, with overweight/underweight positions aver-
aging 6-10 percentage points above or below their bench-
mark weights. Over the 1997-2002 period, however, the
appetite to take on greater currency exposures appears to
have fallen sharply. At present, the average overweight/
underweight currency exposure that investors are main-
taining relative to their benchmark index amounts to just
two or three percentage points relative to market norms
as investors have chosen to position their portfolio closer
to their benchmarks.
What accounts for this drop-off in investors' currency-risk
appetite? We believe that investors have been caught in a
bind in terms of weighting the conflicting forces of valua-
tion and trend. For instance, investors may have felt com-
pelled to underweight the dollar and overweight the euro,
given the dollar's huge overvaluation and the euros huge
undervaluation. But trend-following models have been ar-
guing that investors should take the opposite position.
Given this conflict between valuation and trend, investors
have opted to take on neutral currency exposures and have
been unwilling to shift from this stance for the past few
years. Perhaps investors are waiting for a more definitive
sign that the dollar's long-term appreciating trend has con-
vincingly reversed.
Deutsche Bank Currency-Exposure Index
0
2
4
6
8
10
12
14
1996 1997 1998 1999 2000 2001 2002
(Composite Exposure Index)
Source: Based on the Russell Mellon Survey
Currency Exposure and the Decline in Investor Appetite for FX Risk
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 29
Currency values can be affected by shifts in investors' over-
all appetite to take on risky positions. At any point in time,
investors' overall appetite for risk could lean in favor of
either risk taking or risk aversion. During periods of high
investor confidence, investors tend to increase their expo-
sure to risky assets. During periods of heightened risk aver-
sion, however, investors tend to reduce their exposure to
risky assets. They may close out positions in foreign in-
vestments and return to home base, or they may cut sig-
nificant overweight or underweight exposures relative to
their benchmarks, particularly if such exposures are lever-
aged.
It has been suggested that certain currencies may be more
vulnerable than others during risk-averse periods. For ex-
ample, countries that run large current-account deficits
might see their currencies weaken if foreign capital were
to exit during a period of heightened risk aversion. Like-
wise, high-yield currencies could be vulnerable if inves-
tors sought to unwind risky long high-yield/short low-yield
carry trades in a risk-averse environment.
How does one measure investors' overall appetite for risk?
A number of approaches exist, but by and large they all
seek to capture increases in volatility, wider credit spreads,
trends in commodity prices, and/or shifts in yield curves.
We have constructed the Deutsche Bank Risk Appetite
Index using the components shown in the table below.
There appears to be a strong contemporaneous relation-
ship between the change in the dollar's value and the
change in Deutsche Bank's Risk Appetite Index. As with
the other sentiment/positioning/flow measures, no lead-
ing or predictive relationship is evident, which is to be ex-
pected. After all, when investors’ appetite for risk falls sud-
denly and sharply, it should affect all markets at the same
time. Implied volatility should jump upward, credit spreads
should spike higher, the yield curve should immediately
and perhaps dramatically alter its slope, and currency val-
ues should change at roughly the same time.
This contemporaneous relationship between exchange
rates and risk aversion is evident in the way the Japanese
yen/U.S. dollar exchange rate and the EMBI/U.S. Treasury
yield spread (which is considered to be a measure of
emerging-market risk) have behaved during recent peri-
ods of heightened risk aversion. In the past seven years,
there have been two occasions when the dollar has
tumbled by 20% or more over a very short time span—
January-April 1995 and August-October 1998. In both in-
stances, the EMBI spread exploded to the upside. During
both periods, the EMBI spread and the yen were respond-
ing to heightened risk aversion on a global basis as inves-
tors sought to withdraw from leveraged carry trades that
had suddenly become highly vulnerable.
DB Risk Appetite Index and the U.S. Dollar
-8
-6
-4
-2
0
2
4
6
8
-4
-2
0
2
4
1995 1996 1997 1998 1999 2000 2001 2002
US$ Index(Deviation from Trend)----DB Risk Appetite Index____
Components of the DB Risk Appetite Index
Component Reason for Inclusion
G3 Implied Three-Month FX Volatility Captures FX market uncertainty and risk.
(average of USD/JPY, EUR/JPY
and EUR/USD volatility)
VIX Index Captures uncertainty and risk in the U.S.
(CBOE OEX Volatility Index, and, therefore, global equity markets.
expected volatility
in the S&P100 index)
U.S. High Yield Bond Spread An indicator of risk in U.S. credit markets.
JP EMBI+ Composite Index An indicator of emerging-market volatility,
uncertainty, and risk.
Journal of Commerce Metals Index Proxy for global-growth risks on the
assumption that risk appetites will be
higher during periods of stronger growth.
G3 Yield Curve Proxy for global growth risks, liquidity, and
(10-year bonds less cash rates) safe-haven flows into government bonds.
Construction of the RAI: The DB RAI index is a simple average of the deviation in
each of the above indicators from their trend (standardised). In this way, the index
overcomes problems associated with, for example, the structural widening in credit
spreads in recent years. Periods spent in Risk Taking or Risk Aversion territory have
been determined from the Risk Appetite Index using a standard MACD technique,
designed to pick up shifts in the trend in the RAI.
Source: DB, Datastream
Risk Appetite Shifts and Currency Trends
Emerging-Market Credit Risk & the U.S. Dollar
(1994-1998)
-20
-10
0
10
20
30
200
400
600
800
1000
1200
1400
1600
1800
2000
1994 1995 1996 1997 1998 1999 2000 2001 2002
Yen/US$(3-mo. % chg.)----EMBI Yield Spread(bps.)____
Source: Bloomberg, Datastream
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
30 Deutsche Bank Foreign Exchange Research
Researchers have found that carry trades (long positions
in high-yield currencies funded by short positions in low-
yield currencies) work best in benign risk-appetite envi-
ronments. That is, when investors have a strong appetite
to take on risk, they are more apt to engage in riskier strat-
egies such as carry trades, and if their level of confidence
is high, they may engage in such trades in a leveraged
manner. In contrast, during periods of high risk aversion,
investors are more likely to cut risky positions, particularly
positions that might be highly leveraged. Thus, if the incli-
nation to take on carry trades tends to move positively
with investors’ appetite to take on risk, we would expect
to find a close relationship between the profitability of carry
trades and the trend in risk appetite indicators.
Indeed, as the chart below reveals, the rolling returns from
being long the three highest yielding currencies in the in-
dustrial world and being short the three lowest yielding
currencies in the industrial world closely tracks the trend
in the Deutsche Bank Risk Appetite index. Given this tight
fit, fund managers who undertake carry trades on a fre-
quent basis might find it useful to initiate carry trades when
the risk appetite index is trading upward, and to cut such
positions when the risk appetite index is trending down-
ward.
Risk Appetite and the Profitability of Carry Trades
Deutsche Bank Risk Appetite Index
and the Profitability of Carry Trades
-10
-5
0
5
10
15
20
-15
-10
-5
0
5
10
1998 1999 2000 2001 2002
Fwd.-Bias Strategy Return(%)----DB RAI(index)____
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 31
Exchange-Rate Determination in the
Long Run
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
32 Deutsche Bank Foreign Exchange Research
In the long run, a currency’s value should gravitate toward
its real long-run equilibrium value. If we were able to esti-
mate which exchange-rate level represented long-run fair
value, we would be able to identify the likely path that an
exchange rate should take on a long-term basis. Unfortu-
nately, there is no uniform agreement among economists
on what exchange-rate level represents a currency's true
long-run equilibrium level. The purchasing power parity
(PPP) approach to assessing long-run fair value probably
has the widest following among economists and strate-
gists, but it is also widely recognized that the PPP approach
has serious limitations. That is because historically, the
deviations from estimated PPP values have been both large
and persistent, which suggests that fundamental forces
other than relative national inflation rates have been play-
ing an important role in driving the long-term path of ex-
change rates.
The failure of PPP to hold over medium-term and, in some
cases, long-term horizons has led economists to consider
alternative approaches to assess long-term value in the
FX markets. For instance, the IMF favors the macroeco-
nomic-balance approach to long-term exchange-rate deter-
mination, in which the long-run equilibrium exchange rate
is defined as the rate that would equalize a country's sav-
ings-investment balance with its underlying current-account
balance. According to this approach, if there is a shift in a
country's national savings, investment, or underlying cur-
rent account, then the real long-run equilibrium exchange
rate should adjust accordingly. This approach is quite simi-
lar to the Fundamental Equilibrium Exchange Rate (FEER)
approach pioneered by John Williamson.
More recently, economists have sought to identify the long-
run equilibrium paths of currencies with econometric mod-
els, relating key economic variables to long-term trends in
exchange rates. These studies have found that long-term
trends in relative productivity growth, sectoral-productiv-
ity differentials, persistent trends in a country’s terms of
trade, the net external investment position of a country as
a percentage of its GDP, and trends in national savings and
investment have had some success in explaining the long-
term path of exchange rates.
A Stylized Model of the
Long-Term Trend in a Currency’s Value
Real Exchange Rate
Time
Long-Run
Equilibrium
Path
Exchange-Rate Determination in the Long Run
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 33
The purchasing power parity (PPP) approach to exchange-
rate determination contends that the long-run equilibrium
value of a currency is completely determined by the ratio
of domestic prices relative to foreign prices. PPP is built
on the notion of arbitrage across all tradable goods and
services. Arbitrage ensures that the prices of similar goods
in two countries will be equalized, when expressed in ei-
ther local or foreign currency terms. For example, if the
U.S. and Euroland produced a similar basket of goods, ar-
bitrage should ensure that the prices of those goods would
be the same in both regions. If the price of the U.S. basket
suddenly rose relative to the price of the Euroland basket,
the dollar would need to fall to offset the relative price
change in order to keep the prices of the two baskets equal-
ized.
The evidence on PPP suggests that there are often signifi-
cant and persistent departures from PPP in the short and
medium run. One key reason is that arbitrage in tradable
goods might be limited by transaction costs, tariffs, and
entry and exit barriers. If such barriers to arbitrage were
significant, a large "zone of inaction" might exist whereby
firms would be unwilling to arbitrage traded-goods prices,
and thus significant departures from PPP would exist over
a certain range of exchange-rate movements. Real ex-
change rates would fluctuate randomly inside this zone of
inaction, but if the real exchange rate moved outside of
the transaction bands, arbitrage would finally become prof-
itable enough to bring the real exchange rate back inside
the zone of inaction.
According to a recent survey, U.K.-based foreign-exchange
dealers believe that PPP can be a useful tool for long-run
currency forecasting, but view it as a poor forecasting tool
for the short and medium term. But even for long-run fore-
casting, there does not appear to be overwhelming sup-
port for PPP. According to the survey, only 44.3% believed
that PPP could be useful in predicting exchange-rate move-
ments beyond six months, while 34.9% believed it was
not useful, and 20.8% had no opinion. Survey of U.K.-Based FX Dealers
on the Validity of PPP as a Forecasting Tool
Question: Do you think PPP can be used to predict
exchange-rate movements?
Intraday Medium Run Long Run
(within 6 Months) (over 6 months)
Yes 4.8 16.4 44.3
No 87.4 67.3 34.9
No Opinion 7.8 16.3 20.8
Source: Cheung, Chinn and Marsh, NBER Working Paper 7524, Feb. 2000.
Purchasing Power Parity
Exchange Rate Changes Are Posited to Reflect
Changes in Relative Prices
Relative Price Level
(P /P )
GUS
1.0
DM2000/
US2000
Exchange Rate
(DM/US$)
2.0
DM2000/
US1000
PPP
45 degrees
Purchasing Power Parity
Source: Rosenberg (1996)
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
34 Deutsche Bank Foreign Exchange Research
According to the relative version of PPP, the exchange rate
over time will move to offset differences in national infla-
tion rates. If, for example, the U.S. inflation rate averaged
5% higher than the Euroland inflation rate, the dollar would
need to fall by 5% per annum versus the euro to offset the
gap in relative inflation rates. Thus, if PPP held at all times,
we should expect the nominal exchange-rate path to par-
allel the inflation-differential path.
A country's real exchange rate can be defined as the nomi-
nal exchange rate adjusted for changes in relative inflation
rates. The dollar's real value would be unchanged if the
dollar's nominal value moved exactly in line with changes
in U.S./foreign inflation differentials. Thus, if the U.S. in-
flated at a rate of 5% per annum above the Euroland infla-
tion rate, and the dollar fell by 5% per annum versus the
euro at the same time, the dollar's real value versus the
euro would be unchanged over time. Indeed, if PPP holds
over time, then the real exchange rate would be constant
over time as well.
Many analysts use the trend in real exchange rates as a
gauge to assess a country’s long-run competitiveness. If a
country inflated at a rapid rate relative to its trading part-
ners, and its currency failed to decline in tandem, the
currency's real value would, as a consequence, rise. The
rise in the real exchange rate could be viewed as a mea-
sure of the loss in that country's competitiveness versus
its trading partners. If, instead, the nominal exchange rate
declined to exactly offset the difference in relative infla-
tion rates, then the real exchange rate would have been
constant and there would have been no change in trade
competitiveness.
Real Exchange Rate Determination
q = e - (p* - p)
where:
q = change in the real exchange rate
e = change in the nominal exchange rate
(p* - p) = change in relative price levels
Purchasing Power Parity and the Real Exchange Rate
Relative Price Levels and
Nominal and Real Exchange Rates
Relative Price
Level
(P /P )
EUS
(P /P )
EUS
Time
e
$
U.S. Dollars
Value
(e )
$
Time
Real
Exchange Rate
(q)
qq
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 35
PPP misalignments occur when nominal exchange rates
fail to move in line with relative inflation rates. Consider
what would happen if the U.S. inflation rate averaged 5%
per annum faster than Euroland’s inflation rate but the U.S.
dollar/euro exchange rate was fixed. By preventing a de-
cline in the dollar's nominal value that would offset the
relatively higher U.S. inflation rate, the dollar's real value
q$ = e$ - (pE - pUS)
would be pushed steadily higher as the gap between the
fixed nominal value for the dollar, e$, and a falling inflation
differential, (pE - pUS), steadily widened. The more the U.S.
inflates relative to Euroland, with no offsetting dollar de-
preciation, the less competitive the U.S. would be.
PPP misalignments can also occur if the nominal exchange
rate rises or falls by large amounts relative to modestly
changing or unchanging inflation differentials. For example,
if the U.S. and Euroland inflated at similar rates, but the
dollar's nominal value nevertheless soared versus the euro,
then the dollar's real value
q$ = e$ - (pE - pUS)
would rise as the gap between a rising dollar, e$, and an
unchanged inflation differential, (pE - pUS), steadily widened.
The consequences of a steadily rising real value for the
dollar would be a long-run decline in U.S. competitiveness.
Real Appreciations and Exchange-Rate Misalignments
The Response of Real Exchange Rates
to Changes in Relative Prices
When Nominal Exchange Rates are Fixed
Relative Price
Level
(P - /P )
E- US
(p /p )
EUS
Time
e
$
U.S Dollar’s
Nominal Value
(e )
$
Time
Real
Exchange Rate
(q )
$
q
$
q
$
The Response of Real Exchange Rates
to Changes in Nominal Exchange Rates
When Relative Prices Levels are Stable
Relative Price
Level
(P -/P )
EUS
(p /p )
EUS
Time
e
$
U.S. Dollar’s
Nominal Value
(e )
$
Time
Real
Exchange Rate
(q )
$
q
$
q
$
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
36 Deutsche Bank Foreign Exchange Research
Evidence for both developed and developing countries
suggests that PPP misalignments tend to build up gradu-
ally over time. A key reason why this pattern tends to re-
cur so often is that in the early years of an overly appreci-
ated currency, firms and households find a way to adjust,
and this, at least temporarily, works to temper the harsh
effects of real currency appreciation on the domestic
economy. For example, firms may try to maintain market
share by cutting their profit margins, while households may
attempt to maintain their spending by running down their
savings. Because of these and other adjustments, real
growth might not show any visible signs of weakness and
the current-account balance might not deteriorate signifi-
cantly in the early years of a misaligned exchange rate.
Since an overly appreciated currency is not likely to give
rise initially to a deteriorating trend in domestic economic
activity or in the current-account balance, it is highly un-
likely that the market will feel compelled to wage an attack
on the overvalued currency in the early years of real ap-
preciation. The larger a PPP misalignment grows and the
longer it lasts, however, the greater the likelihood that it
will eventually lead to a visible deterioration in economic
fundamentals. Only when economic fundamentals dete-
riorate beyond a certain threshold level (point
A
in the dia-
gram below) will market participants become emboldened
to wage an attack on the misaligned currency and push it
back toward its real long-run equilibrium level.
PPP Misalignments and Their Eventual Corrections
PPP
Time
% Overvalued
(0)
% Undervalued
(+)
(-)
Weighted-Average Trend
in Fundamentals
Time
Real Exchange Rate
Economic Fundamentals
A
PPP Cycle of Exchange-Rate Misalignment and Correction
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 37
Historically, PPP misalignments exceeding +/- 20% have
been rare and short-lived in the industrial world. When-
ever the DM/US$ exchange rate has risen or fallen by more
than 20% relative to its PPP equilibrium level, the result-
ing misalignment has proven to be transitory.
For the past two years, the dollar has been overvalued by
25%-30%, which is the largest PPP misalignment that the
dollar has experienced since the 1984-85 dollar bubble.
This suggests that the dollar has been in an upside over-
shoot, and since exchange rate misalignments of this size
tend not to be sustainable, one should expect that the
DM/US$ exchange rate will eventually move back inside
its +/- 20% PPP band.
There are signs that the overvalued dollar has been taking
its toll on U.S. competitiveness since 2000. A recent sur-
vey conducted by the Federal Reserve Bank of Philadel-
phia suggests that over 45% of U.S. firms have been nega-
tively affected in some manner by the dollar's overvalua-
tion in the past two years. Further evidence can be found
in the rising chorus of complaints by U.S. industry, as re-
flected in the letter, reprinted below, from leading manu-
facturing associations to the U.S. Treasury Secretary.
The Honorable Paul O’Neill
Secretary of the Treasury
Washington DC 20220
June 4, 2001
Dear Mr. Secretary:
We are writing to tell you that at current levels the exchange value of the dollar is having a strong
negative impact on manufacturing exports, production, and employment. A growing number of American
factory workers are now being laid off principally because the dollar is pricing our products out of
markets – both at home and abroad. Small firms are being affected as well as large ones. As you
balance your responsibilities for international monetary stability and domestic economic growth, we
ask that you take into account the growing burden an overvalued dollar is imposing on U.S. manufacturing.
Since early 1997 the dollar has appreciated by 27 percent. Industries such as aircraft, automobiles and
parts, paper and forest products, machine tools, medical equipment, steel, and other capital goods –
as well as consumer goods producers — are being affected very significantly. No amount of cost
cutting can offset a nearly 30 percent dollar markup.
The total effect on the U.S. economy is staggering. These output losses are particularly serious at this
time, as they coincide with a general economic slowdown. The economic fundamentals have changed
dramatically in the last six months. Production and profitability are down, and manufacturing employment
has fallen by more than a half million jobs since mid-2000. Yet, in the face of slowing economic growth,
declining interest rates, and rising manufacturing unemployment, the dollar has remained high.
In our view, a clarification of Treasury policy is in order, to be certain that it is not seen as endorsing an
ever stronger dollar irrespective of the economic fundamentals. We urge the Treasury to make it clear
that the value of the dollar should be consistent with economic reality and market conditions. This
policy should be buttressed by a commitment to further reductions in interest rates and to cooperating
in exchange markets as appropriate. Moreover, it is vital that the Treasury not condone currency
manipulation by trading partners seeking to make their exports more competitive.
Mr. Secretary, 18 million workers and their families depend directly on the continued strength and
competitiveness of American manufacturing. Many more Americans rely on stockholdings in our
companies for their retirement income. We would like to meet with you to describe more fully the
effects the value of the dollar is having on us. We hope you will be able to accommodate our request.
Respectfully,
Jerry J. Jasinowski, President John W. Douglass, President and CEO
Aerospace Industries Association National Association of Manufacturers
W. Henson Moore, President and CEO Don Carlson, President
American Forest and Paper Association AMT – The Association for Manufacturing Technology
Stephen Collins, President Christopher M. Bates, President & CEO
Automotive Trade Policy Council Motor Equipment Manufacturers Association
The Effect of the Dollar’s Rise on U.S. Business
(Philadelphia Fed Survey)
Question: "Which of the following best
characterizes the effect of the rise in the
value of the dollar on your business?"
Substantial Negative--18.1%
Some Negative Effect--27.7%
No Effect--43.6%
Some Positive Effect--10.6%
Substanial Positive--0.0%
PPP Case Study: Dollar Overvaluation in 2001-02
Source: Datastream
Source: Federal Reserve
Bank of Philadelphia
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
38 Deutsche Bank Foreign Exchange Research
Different approaches to calculating PPP are likely to yield
different estimates of the dollar's PPP fair value. We favor
a long-run averaging approach that allows us to circum-
vent the "base period" problem, in which PPP estimates
can be distorted upward or downward if a particular year is
chosen as the base period. We currently find the dollar to
be significantly overvalued—by over 20% versus the
euro—using either relative consumer or producer price
inflation rates to generate long-run average PPP estimates.
The Economist Magazine's Big Mac Index currently esti-
mates that the dollar is only 10% overvalued versus the
euro. The price of a McDonald's Big Mac hamburger in
Germany and France is roughly equal to the price of the
Big Mac in the U.S., but in Italy and Spain, Big Macs are
20% cheaper than in the U.S.
March 15, 2002 PPP % Over/
Currency (vs. US$) Spot Rate Estimate Under-Valued
Swedish Krona 10.32 7.02 -47.0
Canadian Dollar 1.59 1.24 -27.7
Australian Dollar 0.53 0.72 -27.5
N.Z. Dollar 0.44 0.60 -27.2
Euro 0.88 1.18 -25.4
Norwegian Krone 8.78 7.11 -23.5
Deutsche Mark 2.21 1.80 -22.9
Danish Krone 8.41 7.07 -18.9
Swiss Franc 1.66 1.45 -14.5
Japanese Yen 129.06 117.32 -10.0
British Pound 1.43 1.55 -8.0
US$ Index 117.18 96.32 21.7
CPI-Based Purchasing Power Parity (PPP) Estimates March 15, 2002 PPP % Over/
Currency (vs. US$) Spot Rate Estimate Under-Valued
Swedish Krona 10.32 7.73 -33.4
Euro 0.88 1.14 -22.7
Australian Dollar 0.53 0.67 -22.1
Swiss Franc 1.66 1.38 -20.0
Deutsche Mark 2.21 1.85 -19.8
N.Z. Dollar 0.44 0.52 -15.9
Norwegian Krone 8.78 7.75 -13.3
Danish Krone 8.41 7.46 -12.8
Canadian Dollar 1.59 1.43 -11.2
Japanese Yen 129.06 117.36 -10.0
British Pound 1.43 1.46 -2.6
US$ Index 117.18 104.22 12.4
PPI-Based Purchasing Power Parity (PPP) Estimates
Big Mac PPP Estimates
(Economist Magazine’s Big Mac Index)
-60
-40
-20
0
20
40
60
Switz.
Denmark
Britain
U.S.
Argentina
France
Japan
Mexico
Sweden
Germany
Euro area
S. Korea
Canada
Taiwan
Chile
Spain
Italy
Singapore
Brazil
Australia
Poland
N.Z.
Czech Rep.
Hong Kong
Indonesia
Hungary
Russia
Thailand
China
S. Africa
Malaysia
Philippines
%
Over(+) Under(-) Valued
U.S. Dollar PPP Estimates in 2002
Source: www.economist.com
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 39
PPP estimates are often better guides to medium/long-
run exchange-rate movements when traded-goods price
indices are used to derive PPP estimates rather than when
non-tradable goods price indices are used. For example,
the yen tends to move much more in line with the trend in
Japanese/U.S. relative export prices than it does with the
trend in relative consumer prices.
Since PPP operates on the principle of arbitrage, interna-
tional trade should ensure that tradable goods prices in
different countries are broadly similar when measured in a
single national currency.
The Yen and Relative Unit Labor Costs
0.80
0.90
1.00
1.10
1.20
1.30
1.40
1.50
50
100
150
200
250
300
350
76 78 80 82 84 86 88 90 92 94 96 98 00 02
Japan/U.S. ULCs----
Y
en/US$____
The Yen and Relative Producer Prices
0.60
0.80
1.00
1.20
1.40
1.60
1.80
50
100
150
200
250
300
350
76 78 80 82 84 86 88 90 92 94 96 98 00 02
Japan/U.S. Wholesale Prices----
Y
en/US$____
The Yen and Relative Consumer Prices
0.50
0.60
0.70
0.80
0.90
1.00
1.10
1.20
50
100
150
200
250
300
350
76 78 80 82 84 86 88 90 92 94 96 98 00 02
Japan/U.S. Consumer Prices----
Y
en/US$____
The Yen and Relative Export Prices
0.20
0.40
0.60
0.80
1.00
1.20
50
100
150
200
250
300
350
76 78 80 82 84 86 88 90 92 94 96 98 00 02
Japan/U.S. Export Prices----
Y
en/US$____
Purchasing Power Parity and the Yen
Source: Datastream Source: Datastream
Source: Datastream Source: Datastream
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
40 Deutsche Bank Foreign Exchange Research
There exists a huge volume of empirical work on whether
or not PPP represents a valid tool for forecasting purposes.
The weight of evidence suggests that although there are
often significant and persistent departures from PPP in both
the short and medium term, exchange rates do exhibit a
tendency to gravitate toward their PPP values in the long
run.
The consensus among academic researchers is that the
speed of convergence to PPP is quite slow—PPP devia-
tions appear to dampen at a rate of roughly 15% per year.
Empirical Evidence on PPP
This places the half-life of PPP deviations at around 4½
years for exchange rates in industrial nations. In other
words, for any given deviation of an exchange rate from its
estimated PPP value, roughly half of that deviation should
be removed in 4½ years' time. As the charts below show,
there is no positive relationship between changes in ex-
change rates and changes in relative inflation rates on a
one-year basis. However, as the time horizon is length-
ened, to six years and beyond, a strong positive relation-
ship becomes apparent.
The Impact of Relative Inflation Rates on Exchange Rates
Over Different Time Horizons
Adapted from Peter Isard, Hamid Faruqee, G. Russell Kincaid, and Marin Fetherston,
“Methodology for Current Account and Exchange Rate Assessments”,
IMF Occasional Paper 209, 2001.
50
40
30
20
10
0
-10
-20
-30
-40
-50 -50 -40 -30 -20 -10 0 10 20 30 40 50
50
40
30
20
10
0
-10
-20
-30
-40
-50 -50 -40 -30 -20 -10 0 10 20 30 40 50
50
40
30
20
10
0
-10
-20
-30
-40
-50 -50 -40 -30 -20 -10 0 10 20 30 40 50
50
40
30
20
10
0
-10
-20
-30
-40
-50 -50 -40 -30 -20 -10 0 10 20 30 40 50
% Change in Exchange Rate % Change in Exchange Rate
% Change in Exchange Rate % Change in Exchange Rate
% Change in
Relative Inflation
% Change in
Relative Inflation % Change in
Relative Inflation
% Change in
Relative Inflation
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 41
PPP and Investment Strategy
Source: Datastream, Bloomberg
Source: Datastream, Bloomberg
Since many empirical studies find that exchange rates ex-
hibit a tendency to gravitate toward their PPP levels in the
long run, long-term investors and issuers would stand to
benefit by positioning themselves for an eventual correc-
tion of a large PPP misalignment. If investors could count
on a rapid return to PPP, they could purchase undervalued
currencies in the hope that the undervaluation would soon
be corrected via an appreciation of the currency. Similarly,
they would look to sell overvalued currencies.
For instance, the euro's gross PPP undervaluations ver-
sus both the dollar and the yen do not appear to be sus-
tainable, given the long-run track record of these two
crossrates. Yet long-dated euro/dollar and euro/yen forward
exchange rates are not anticipating a significant correction
in the euro’s PPP misalignment over the next five years.
With conventional PPP analysis arguing for a long-term ap-
preciation of the euro versus both the dollar and the yen,
this would seem to present an opportunity for investors
and issuers to exploit the divergent trends in PPP and long-
dated forward rates.
Because the deviations from PPP can be large and persis-
tent over short and medium-term horizons, it can be dan-
gerous to pursue strategies that are based solely on an
expected rapid return of exchange rates to their estimated
PPP values. The major risk to such strategies is that the
deviations from PPP may widen further instead of narrow-
ing. Investors who embrace a PPP approach to investment
strategy would have to have substantial risk capital on hand
to absorb losses that could be incurred if the departures
from PPP proved to be sizable and persistent.
One way to manage FX risk using a PPP approach to in-
vestment strategy would be to gradually take on long po-
sitions in undervalued currencies (or short positions in over-
valued currencies) as misalignments build. Modest posi-
tions would be undertaken when the misalignments were
small, and if the deviations from PPP grow and persist,
more aggressive positions could then be undertaken.
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
42 Deutsche Bank Foreign Exchange Research
According to the macroeconomic-balance approach to long-
run exchange-rate determination, a currency's real long-
run equilibrium value is defined as that rate that equalizes
a country's savings-investment balance with its underly-
ing current-account balance. The IMF uses the macroeco-
nomic-balance approach to derive quantitative estimates
of long-run fair value for all industrial-country currencies.
The IMF's approach is a three-step procedure.
1. The IMF estimates a trade-equation model to calcu-
late a country's underlying current-account position.
2. The IMF then estimates a national income model to
determine the normal or sustainable savings-invest-
ment imbalance that would prevail over the medium
run if that country were operating at its full-employ-
ment potential.
3. Finally, the IMF estimates the equilibrium exchange
rate that would equate the underlying current-account
imbalance with the sustainable savings-investment
gap.
A country's underlying current-account balance tends to
move inversely with the real exchange rate. That is, a
country's current account would tend to improve as the
currency's real value moved lower and its tradable goods
became more competitive. Hence, the underlying current-
account balance, CAB, is shown to be negatively related
to the real exchange rate in the top diagram.
A country's savings-investment gap typically moves posi-
tively with the real exchange rate. Although the overall level
of national savings does not typically respond to changes
in the exchange rate, investment spending tends to move
inversely with changes in the real exchange rate. This is
because a rising real exchange rate lowers profitability,
which in turn discourages investment spending. Thus, as
the real exchange rate rises, investment spending should
fall, and the gap between savings and investment should
therefore rise. Hence, the savings-investment, S-I, gap is
shown in the middle diagram to move positively with the
real exchange rate.
A currency's real long-run equilibrium value is determined
at the point where a country's savings-investment imbal-
ance just matches its underlying current-account imbal-
ance. The equilibrium exchange rate is shown to be q0 in
the bottom diagram where the S-I and CAB schedules in-
tersect at point
E
.
The Macroeconomic-Balance Approach
to Long-Run Exchange-Rate Determination
The Underlying Current-Account Balance
and the Real Exchange Rate
Current Account Balance
Real Exchange Rate
(-) (0) (+)
q
1
CAB
The Savings-Investment Schedule
and the Real Exchange Rate
Savings-Investment
Real Exchange Rate
(-) (0) (+)
S-I
q
2
The Equilibrium Real Exchange Rate
Is Determined When the Current-Account Gap
Equals the Savings-Investment Gap
Current-Account Balance,
Savings-Investment
Real Exchange Rate
(-) (0) (+)
q
1
S-I
q
0
q
2
CAB
E
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 43
If a currency's real long-run equilibrium value is determined
by the interaction of savings, investment, and external-
balance considerations, its value would be constant over
time only if there were no structural changes on the sav-
ings, investment, or external-balance fronts over the rel-
evant time period. (This is illustrated in the top diagram.)
When structural changes on those three fronts occur, how-
ever, the real long-run equilibrium exchange rate will ei-
ther rise or fall in response to those changes. A structural
rise in investment, a sustained move toward fiscal stimu-
lus, or a persistent decline in private savings should give
rise to a trend increase in the real long-run equilibrium ex-
change rate over time, as shown in the bottom diagram.
The opposite would give rise to a trend decline in the real
long-run equilibrium exchange rate.
Equilibrium Exchange Rate—Constant or Variable?
Investment-Spending
Schedule Stable Over Time
Savings minus Investment,
Current-Account Surplus
Equilibrium Real
Exchange Rate (q)
CAB
1
q
1
(0)
S-I
1
Current-Account Deficit
Structural
Shifts in
Investment
Spending
Savings minus Investment,
Current-Account Surplus
Equilibrium Real
Exchange Rate (q)
S-I
2
CAB
1
q
2
q
1
(0)
S-I
1
Current-Account Deficit
S-I
3
S-I
4
q
3
q
4
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
44 Deutsche Bank Foreign Exchange Research
Structural changes on a country's
internal-balance
front can
exert a powerful influence on the trend in exchange rates
over time. When a country enjoys a major investment
boom, such as the U.S. experienced in the second half of
the 1990s, it will give rise to an upward shift in the S-I
schedule and to a rise in the equilibrium exchange rate
over time. Similarly, a major move toward fiscal stimulus,
as the U.S. experienced in the first half of the 1980s, would
shift the S-I curve upward and push the currency's real
long-run equilibrium value higher, as shown in the top dia-
gram.
Structural changes on a country's
external-balance
front
can also exert a powerful influence on the trend in exchange
rates over time. Productivity changes and technological
innovations in a country's traded-goods sector can give
rise to a sustained improvement in a country's current-
account balance and thereby contribute to a trend rise in
the real exchange rate over time, as shown in the middle
diagram. Japan's large and persistent current-account sur-
pluses played a key role in driving the yen's real value higher
in the 1980s and early 1990s.
During certain periods, structural changes might be occur-
ring on both the internal and external-balance fronts at the
same time. The net impact of those changes will deter-
mine whether the exchange rate will rise or fall. For ex-
ample, in the case of Japan in 2000-01, both internal and
external-balance forces were working jointly to lower the
yen's real long-run equilibrium value. Japan's fiscal stance
was tightening, investment spending was declining, and,
at the same time, Japan's current-account surplus was
undergoing a structural deterioration as competition from
Asia intensified. As illustrated in the bottom diagram, those
forces combined to cause the yens equilibrium value to
decline from q1 to q2.
A Shift in Investment Behavior
Can Give Rise to an Upward Revision
in a Currency’s Long-Run Equilibrium Value
Savings minus Investment,
Current-Account Balance
Real Exchange Rate
Structural
Increase in
Investment
Spending
S-I
2
CAB
1
Upward Revision in the
Currency’s Real Long-Run
Equilibrium Value
q
2
q
1
(0)(-) (+)
S-I
1
A Structural Improvement in Competitiveness Can Give
Rise to an Improvement in the Current-Account Balance
and in a Currency’s Real Long-Run Equilibrium Value
Savings minus Investment,
Current-Account Balance
Real Exchange Rate
Structural
Improvement
in Current
Account
CAB
1
Upward Revision in the
Currency’s Real Long-Run
Equilibrium Value
q
2
q
1
(0)(-) (+)
S-I
1
CAB
2
The Combination of a Structural Decrease in Investment
and a Deterioration in the Current-Account Balance
Will Give Rise to a Downward Revision in
a Currency’s Real Long-Run Equilibrium Value
Savings minus Investment,
Current-Account Balance
Real Exchange Rate
Structural
Deterioration
in Current
Account
CAB
1
Downward Revision in the
Currency’s Real Long-Run
Equilibrium Value
q
2
q
1
(0)(-) (+)
S-I
1
CAB
2
Structural
Decrease in
Investment or
Gov’t Spending
S-I
2
Changes in Real Long-Run Equilibrium Exchange Rates
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 45
The problem for currency forecasters and investment strat-
egists is how to precisely quantify what exchange rate level
truly represents a currency's real long-run equilibrium level.
Estimates of the equilibrium exchange rate are often sen-
sitive to the assumptions that a macro-econometric mod-
eler might make. Indeed, it is quite possible that a wide
range of equilibrium exchange-rate estimates are plausible,
with each estimate dependent on the underlying assump-
tions made.
Estimates of the dollar's equilibrium value by the Institute
for International Economics show the dollar to be signifi-
cantly overvalued versus the yen and euro by 25%-30%.
A recent report by the OECD surveyed a broad range of
econometric studies that constructed estimates for the
euro's long-run equilibrium value. Although the estimates
vary widely, they tend to center around the US$/ 1.10-
1.20 range, suggesting that the euro is significantly under-
valued versus the U.S. dollar.
Selected Estimates of the Euro's
Medium/Long-Run "Equilibrium" Value Equilibrium
Key Explanatory Exchange Rate
Study Variables/Model Estimate
(US$/)
Wren-Lewis & Driver (1998) FEER Model 1.19-1.45
Borowski and Couharde (2000) FEER Model 1.23-1.31
Alberola et al. (1999) Ratio of Nontraded/Traded Goods Prices, 1.26
Net Foreign Assets
Chinn and Alquist (2000) M1, GDP, Short-Term Interest Rates, CPI, 1.19-1.28
Ratio of Nontraded/Traded Goods Prices
Lorenzen and Thygessen (2000) Net Foreign Assets, R&D Spending, 1.17-1.24
Demographics,
Ratio of Nontraded/Traded Goods Prices
Duval (2001) Consumption, Multi-Factor Productivity, 1.15
Real Long-Term Yield Spread,
Ratio of Nontraded/Traded Goods Prices
Clostermann and Schnatz (2000) Real Long-Term Yield Spread, Oil Price, 1.13
Government Spending,
Ratio of Nontraded/Traded Goods Prices
Teiletche (2000) Productivity, Government Spending, 1.09
Real Long-Term Yield Spread, M1,
Industrial Production
OECD GDP PPP 1.09
Gern et al. Short-Term Real Interest-Rate Differential 1.03
Schulmeister PPP for Tradeables 0.87
Deutsche Bank PPP (Long-Run Average) 1.20
Source: OECD Working Paper Number 298, June 2001 (for other than Deutsche Bank).
Equilibrium Exchange Rate Estimates
Fundamental Equilibrium Exchange Rate Estimates
(Estimates of U.S. Dollar FEERs for 2000)
Euro Japanese Yen
Range 1.135-1.387 77.3-94.5
Midpoint 1.26 86.0
Source: Simon Wren-Lewis and Rebecca Driver,
Exchange Rates for the Year 2000
, Institute for International Economics, 1998.
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
46 Deutsche Bank Foreign Exchange Research
Many economists cite the rise in U.S. productivity growth
relative to productivity growth overseas beginning in the
second half of the 1990s as one of the key variables that
drove the dollar sharply higher versus most major curren-
cies. Clearly, the gains in productivity were a uniquely U.S.
phenomenon. A recent Federal Reserve Board Study found
that multifactor productivity growth in the U.S. surged in
the second half of the 1990s, while it decelerated in Japan
and was broadly unchanged in Euroland.
Fed Chairman Alan Greenspan has been a strong advo-
cate of the thesis that the dollar’s seven-year uptrend was
productivity-related. According to Mr. Greenspan, "The evi-
dent strengthened demand for the dollar, relative to the
euro, has reflected a market expectation that productivity
growth in the United States is likely to be greater than in
continental Europe in the years ahead."
The importance of relative productivity growth as a key
determinant of the dollar's value versus the euro can be
gleaned from the accompanying charts. U.S. and Euroland
labor productivity growth did not differ by much in the 1980s
and early 1990s. But beginning in 1995, U.S. productivity
growth surged relative to the trend in Euroland. Indeed,
the trend in long-term Euroland/U.S. relative productivity
growth appears to explain a large portion of the 1995-2000
slide in the euro's value versus the U.S. dollar.
U.S. and Euroland Productivity
(Derived from OECD Calculations of
Unit Labor Costs and Compensation Growth Rates)
0
500
1000
1500
2000
2500
3000
1983 1985 1987 1989 1991 1993 1995 1997 1999 2001
Euroland----
(Index) U.S.____
Source: OECD Economic Outlook
Productivity Growth Rates
(Percentage Changes in Multifactor Productivity)
1990-95 1996-99
(%) (%)
U.S. 0.79 1.47
Canada 0.26 0.27
France 0.89 1.12
Germany 1.02 1.07
Italy 1.32 -0.14
Japan 1.31 0.85
U.K. 1.21 0.95
Australia 1.15 2.11
Denmark 2.37 0.31
Norway 2.48 1.13
Switzerland -0.57 0.84
Source: Gust, C. and Marquez, J., “Productivity Developments Abroad”,
Federal Reserve Bulletin
, October 2000.
U.S./Euroland Productivity Ratio
and the U.S. Dollar/Euro Exchange Rate
0.90
1.00
1.10
1.20
1.30
1.40
0.6
0.7
0.8
0.9
1.0
1.1
1.2
87 88 89 90 91 92 93 94 95 96 97 98 99 00
US$/Euro----Euroland/US Productivity(ratio)____
Source: OECD Economic Outlook
Productivity Trends and Exchange Rates
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 47
There has been a great deal of debate among economists
over what contributed to the recent stellar gains in U.S.
productivity and whether those gains will prove sustain-
able in the long term. A number of researchers believe
that the absolute gains in U.S. productivity were largely an
information technology (IT) phenomenon, and since the IT
boom was largely a U.S. story, the relative gains in U.S.
productivity could be explained by IT as well. Many of those
same researchers believe that the recent gains in U.S. la-
bor productivity will prove sustainable in the long run, which
would be dollar positive. On the other side of the argu-
ment, some researchers are not convinced that the U.S.
enjoyed a productivity miracle, contending instead that
favorable cyclical factors might have exaggerated the rise
in reported productivity. If that were the case, then U.S.
productivity growth would be expected to fall back to its
long-term trend.
A recent Conference Board report demonstrates that the
relatively robust U.S. productivity gains were due not only
to the comparative advantage of IT firms in the U.S., but
also to the ability of U.S. non-IT firms to use the new infor-
mation technologies developed by the IT sector to boost
their overall productivity. This was made possible by U.S.
firms' willingness to invest in these new technologies. In
contrast, non-IT firms in Euroland and Japan were slow to
invest in new technologies, even though such technolo-
gies were as available to them as they were to non-IT firms
in the U.S. The net result is that productivity gains in the
non-IT sector of the U.S. were several times greater than
the productivity gains by comparable firms in Euroland and
Japan.
According to the Conference Board report, the failure of
Euroland firms to embrace the new technologies "can be
traced to insufficient liberalization and impediments to
market evolution." The report adds that Euroland "barriers
to change…are widespread….and opportunities are still
restricted in many information and communications tech-
nology-using industries, such as transportation, communi-
cation and banking." The report emphasized that Euroland
"regulatory rigidities inhibit reallocations of labor and capi-
tal to their most productive uses, reducing the benefits to
be obtained from investment in new technologies."
The Conference Board holds out the hope that "new com-
petitive rules in communications, banking reforms, and
even the adoption of the single currency" may make it pos-
sible for Euroland to experience stronger productivity gains
in the future, which would be positive for the euro. On a
negative note, however, the report argues that "recent com-
pilations of economic and administrative regulations sug-
gest that Europe and Japan still have some way to go be-
fore they will be able to fully exploit the new computer and
information technologies. Failure to move forward on regu-
latory liberalization and the reduction of impediments to
market evolution would perpetuate the present under-in-
vestment in information and communications technology."
The report then wryly concludes that "realizing the poten-
tial growth effects from information and communications
technology is not automatic."
One implication of the Conference Board report is that U.S.
firms may enjoy stronger productivity gains than their coun-
terparts in Euroland and Japan for a while longer. This sug-
gests that until Euroland begins to close the productivity
gap with the U.S. on a sustained basis, the euro will con-
tinue to encounter difficulty in making significant headway
versus the dollar.
Contribution of IT-Producing Industries to
Labor Productivity Growth in G-7 Countries
(1995-1999)
0.30.3
0.40.40.4
0.6
0.7
0.0
0.2
0.4
0.6
0.8
U.S. U.K. Japan Germany France Italy Canada
Percentage Contribution(%)
Source: Conference Board
Contribution of IT-Using Industries to
Labor Productivity Growth in G-7 Countries
(1995-1999)
0.4
0.20.2
0.5
0.3
0.5
1.4
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
U.S. U.K. Japan Germany France Italy Canada
Percentage Contribution(%)
Source: Conference Board
U.S. Productivity Growth Surge and the Dollar—A Medium or
Long-Term Phenomenon?
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
48 Deutsche Bank Foreign Exchange Research
A new study by McKinsey & Company, Inc. ("U.S. Produc-
tivity Growth: 1995-2000") sheds additional light on the
sources of the U.S. productivity surge. The McKinsey study
contends that U.S. productivity increased in the 1995-2000
period for both cyclical and structural reasons. The study
finds that U.S. labor-productivity growth rose from an aver-
age 1.4% per annum over the 1987-95 period to an aver-
age 2.5% per annum over the 1995-2000 period, with only
about one-half of the incremental gains projected to be
long-lasting.
The McKinsey study projects that U.S. labor productivity
growth will ease a bit toward an average 2.0% per annum
pace over the next five years as the influence of shorter-
run cyclical factors washes out. While this would repre-
sent a slowdown from the 1995-2000 pace, it would nev-
ertheless exceed the pace of productivity growth in 1987-
95.
A 0.5% per annum increase in labor-productivity growth
might not seem like a lot, but if such a gain were sus-
tained over a 10-year period, it could contribute up to $1.2
trillion in additional surpluses on the U.S. Federal govern-
ment budget balance. It could also cut the cost of a 50-
year fix to Social Security in half.
The McKinsey study finds that the gain in U.S. labor pro-
ductivity over the 1995-2000 period was more than just an
IT story. The study contends that regulatory changes, strong
competition, better management, and favorable cyclical
factors all played a positive role in boosting U.S. productiv-
ity. Interestingly, the study finds that of the 59 sectors of
the U.S. economy that were investigated, only six sectors,
representing 28% of the economy, contributed to the 1995-
2000 productivity gains. The six sectors that made sub-
stantial contributions to U.S. productivity growth were: 1)
wholesale trade, 2) retail trade, 3) security and commod-
ity brokers, 4) electronic equipment/semiconductor firms,
5) industrial equipment/computer firms, and 6) telecom
services. The other 53 sectors, representing 72% of the
economy, made either small positive or negative contribu-
tions that effectively washed each other out for a net-zero
contribution to the U.S. economy's total productivity gains.
Of concern to both the U.S. economy and the dollar is the
fact that 72% of the overall U.S. economy did not partici-
pate in the productivity gains registered in the past five
years. The dollar rose strongly because the gains regis-
tered by the other 28% were so powerful that they were
able to boost the overall U.S. economy's productivity
growth far above the pace of productivity growth in
Euroland and elsewhere. For those sectors that enjoyed
strong productivity gains, the dollar's rise has not dented
their overall competitiveness, as the positive effects of a
relative rise in productivity have effectively offset the nega-
tive effects of a rising dollar. But the other 72% that did
not participate in the productivity surge have been left to
compete in world markets with an overvalued exchange
rate without an offsetting gain in productivity. That is why
a large number of U.S. firms have begun to complain that
the dollar's strength is negatively affecting their long-run
competitiveness.
1987-95
Productivity
Growth
1995-99
Productivity
Growth
(By Sector)
U.S High- & Low-Tech Industrial Production
(Year-over-Year Percentage Changes)
-20
0
20
40
60
1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002
Total excluding High-Tech----
(%) Computers/Communications/Semiconductors____
U.S. Productivity Gains and the Dollar—Is the Dollar Now Fairly or Over-Valued?
Source: Datastream
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 49
Productivity gains will tend to exert a greater impact on
exchange rates if those productivity gains are concentrated
in the traded-goods sector of an economy. A recent Fed-
eral Reserve Bank of New York study ("To What Extent Does
Productivity Drive the Dollar?",
Current Issues in Econom-
ics and Finance
, Volume 7, Number 8, August 2001) finds
that most of the gains in U.S. productivity in the 1990s
were concentrated in the U.S. traded-goods sector.
Looking at sectoral productivity gaps (the difference be-
tween traded-goods productivity and non-traded-goods pro-
ductivity) in the U.S., Japan, and Euroland, one finds that
the U.S. sectoral productivity gap has risen in the past 30
years as the gap between U.S. traded and non-traded pro-
ductivity growth has widened. In contrast, the sectoral pro-
ductivity gap has fallen in Japan and has been steady in
Euroland. The New York Fed's findings reveal that relative
sectoral productivity gaps explain more than two-thirds of
the movement in the yen and euro in the past decade.
U.S. Productivity Growth by Sector
(1970-1999)
1.2
0.40.4
4.6
3.2
1.0
0
1
2
3
4
5
1970-80 1980-90 1990-99
Nontraded Sector(dark)
(%) Traded Sector(light)
Source: OECD; Federal Reserve Bank of New York,
August 2001
Japan Productivity Growth by Sector
(1970-1999)
0.2
2.3
2.4
1.7
4.7
5.4
0
1
2
3
4
5
6
1970-80 1980-90 1990-99
Nontraded Sector(dark)
(%) Traded Sector(light)
Source: OECD; Federal Reserve Bank of New York,
August 2001
Euroland Productivity Growth by Sector
(1970-1999)
0.9
1.2
1.8
2.9
3.0
4.0
0
1
2
3
4
5
1970-80 1980-90 1990-99
Nontraded Sector(dark)
(%) Traded Sector(light)
Source: OECD; Federal Reserve Bank of New York,
August 2001
U.S. Productivity and Changes in the
U.S./Euro Real Exchange Rate
(1970-1999)
1.1
0.8
-1.0
1.8
0.1
-4.9
-6
-4
-2
0
2
4
1970-80 1980-90 1990-99
HBS Productivity Measure(dark)
(%) Real Exchange Rate Chg.(light)
Note: Harrod, Balassa, Samuelson Productivity.
Source: OECD; Federal Reserve Bank of New York,
August 2001.
U.S. Productivity and Changes in the
U.S./Japan Real Exchange Rate
(1970-1999)
1.5
0.4
-1.6
1.9
0.6
-2.3
-3
-2
-1
0
1
2
3
1970-80 1980-90 1990-99
HBS Productivity Measure(dark)
(%) Real Exchange Rate Chg.(light)
Note: Harrod, Balassa, Samuelson Productivity.
Source: OECD; Federal Reserve Bank of New York,
August 2001.
Productivity Growth in the Traded-Goods Sector and the Determination of
Exchange Rates
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
50 Deutsche Bank Foreign Exchange Research
The Norwegian Krone and the Price of Oil
(since January 1999)
5
10
15
20
25
30
35
40
(4.80)
(4.60)
(4.40)
(4.20)
(4.00)
(3.80)
1999 2000 2001 2002
Brent Oil(US$/bbl.)----Nkr/DM(reverse scale)____
Changes in a country's terms of trade can, at times, exert
a strong influence on the direction that exchange rates
take. For instance, in commodity-oriented industrial econo-
mies such as Australia, New Zealand, and Canada, where
commodity exports make up a large part of GDP (relative
to the rather small shares in the G3 economies), one finds
a strong positive relationship between the trend in relevant
commodity-price indices and the trend in the A$, NZ$, and
C$, respectively.
In the case of the euro, the price of oil has been found to
be an important explanatory variable, with higher oil prices
contributing to a weaker euro, and vice versa. A recent
Bank of Canada study found that a similar relationship held
between energy prices and the C$, with higher energy
prices bearish for the C$ and lower energy prices bullish
for the C$.
A recent study by the central bank of Norway (Norges Bank)
finds that there is a non-linear relationship between oil
prices and the Norwegian krone. A change in oil prices
tends to have a larger impact on the krone's value when
the price of oil is below $14 a barrel or above $20 a barrel.
When oil prices fluctuate inside the $14-$20 range, the
impact of oil price changes on the krone's value is found
to be insignificant.
The Australian Dollar and Commodity Prices
Reserve Bank of Australia Commodity Price Index
70
80
90
100
110
120
0.40
0.50
0.60
0.70
0.80
0.90
1.00
84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02
RBA Commodity Price Index(US$-Terms)----US$/A$____
The Euro and the Price of Oil
(since April 1997)
5
10
15
20
25
30
35
40
(1.30)
(1.20)
(1.10)
(1.00)
(0.90)
(0.80)
1997 1998 1999 2000 2001 2002
Brent Oil(US$/bbl.)----US$/Euro(reverse scale)____
The Canadian Dollar and Commodity Prices
Bank of Canada Commodity (excl. Energy) Index
80
90
100
110
120
130
140
(1.65)
(1.60)
(1.55)
(1.50)
(1.45)
(1.40)
(1.35)
(1.30)
1995 1996 1997 1998 1999 2000 2001 2002
BoC Non-Energy Commod. Price Index----C$/US$(reverse scale)____
Commodity Exports as a Percentage of GDP
(for Selected Industrial Countries)
0.5
1.8
3.7
5.2
12.0
12.1
19.4
0
5
10
15
20
25
N.Z. Canada Australia UK Germany U.S. Japan
(%)
Source: Datastream Source: Datastream
Source: Datastream Source: Datastream
Source: OECD
Terms of Trade Changes and Exchange Rates
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 51
Several researchers have noted that there exists a posi-
tive long-run relationship between a country's net interna-
tional investment position as a percentage of GDP and its
real effective exchange rate. There are several reasons why
this relationship tends to hold.
First, exchange rates typically do not adjust to equilibrate
the current-account balance in each and every period. The
reason for this is that a deficit country may be able to at-
tract sufficient capital inflows to finance its deficit for a
considerable time. Indeed, if those inflows are consider-
able, the deficit country's currency might, in fact, rise even
as the current-account deficit widens—witness the U.S.
in 1981-85 and again in 1995-2002.
If a country's current-account deficit rises persistently over
time, however, there is likely to come a point when it can
no longer be easily financed. When that point is reached, a
currency's real value could then come under considerable
downward pressure. Hence, although the exchange rate
and the current-account imbalance may not appear to be
positively related on a monthly or quarterly basis, cumula-
tive trends in current-account imbalances and exchange
rates tend to be more closely aligned.
A second reason for the positive relationship between the
trend in net international investment positions and the trend
in real exchange rates is the effect of the transfer of wealth
implied by external-account imbalances. The cumulative
trend in current-account imbalances gives rise to shifts in
the residence of wealth from deficit to surplus countries,
and such wealth transfers give rise to portfolio adjustments
that put upward pressure on surplus-country currencies
and downward pressure on deficit-country currencies over
time.
Given that most investors prefer to keep the bulk of their
wealth in domestic and not foreign-currency assets—the
so-called home-country bias—investors in surplus coun-
tries will find themselves accumulating more foreign-cur-
rency assets than they desire relative to their domestic
holdings. Over time, surplus country investors will attempt
to rebalance their portfolios in favor of domestic-currency
assets. As they do, the surplus country's currency will tend
to rise relative to the deficit country's currency. The Japa-
nese yen and the Deutschemark—to a lesser degree—
underwent long-term appreciations from the 1980s until
the mid-1990s as their cumulative current-account sur-
pluses rose.
The Japanese Yen and the U.S./Japan
Cumulative Current-Account/GDP Differential
(50)
(40)
(30)
(20)
(10)
0
10
20
50
100
150
200
250
300
77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96
Cum. Curr.-Acct. as % of GDP Diff.---- (%)
¥
/US$____
The Deutschemark and the U.S./German
Cumulative Current-Account/GDP Differential
(50)
(40)
(30)
(20)
(10)
0
1.00
1.50
2.00
2.50
3.00
3.50
85 86 87 88 89 90 91 92 93 94 95 96
Cum. Curr.-Acct. as % of GDP Diff.---- (%)DM/US$____
Source: Datastream Source: Datastream
Net International Investment and the Equilibrium Exchange Rate
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
52 Deutsche Bank Foreign Exchange Research
A final reason why the trends in net international invest-
ment positions and real exchange rates are positively re-
lated is that heavily indebted nations need to service their
net external debts, and that can only be done if the debtor
country runs a trade surplus to generate the income to
service those debts. In order for the debtor country to run
a trade surplus consistently over time, the real exchange
rate of a debtor country would need to decline. Similarly, a
country with a large cumulative current-account surplus
would need to see its currency's real value rise over time
to insure that the surplus and the associated net foreign
asset position did not continue rising to unsustainably high
levels.
Although the U.S. has seen its net international investment
position deteriorate sharply in the past decade, the dollar
has nevertheless remained extraordinarily resilient. One
explanation is that, although U.S. external liabilities exceed
U.S. external assets by a wide margin, the U.S. has tended
to earn a higher return on its external assets than it has
paid out on its external liabilities. The result has been that
net investment income flows, until recently, posted a mod-
est surplus. While U.S. net investment income flows have
recently swung into a modest deficit position, the deficit
does not appear to be sizeable enough at this stage to
seriously damage the dollar's long-run prospects.
U.S. Balance on Investment Income
(Four-Quarter Moving Sum)
-30
-20
-10
0
10
20
30
40
1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002
(US$ bn.)
U.S. Net Investment Position
(at Current Cost and Market Value)
-2000
-1500
-1000
-500
0
500
80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99
Market Value(light)
(US$ bn.) Current Cost(dark)
Source: Datastream Source: Datastream
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 53
The dollar has exhibited a tendency to both rise and fall
over long-term cycles, with each of the dollar cycles since
the floating of exchange rates in 1973 sharing a number of
common features.
First, the magnitude of the dollar's exchange-rate move-
ments in each cycle has generally been large, often far
exceeding the forecasts of portfolio managers, market pun-
dits and forward-exchange rates.
Second, the period of each dollar cycle has tended to be
long, often surpassing by a wide margin the typical length
of U.S. and foreign business cycles. That would suggest
that cyclical factors alone cannot account for the tendency
of the dollar to move in large and protracted swings.
Third, at the end of each major cycle, there has been a
tendency for the dollar to overshoot its long-run equilib-
rium level, often in a climactic fashion.
Finally, when the dollar has overshot its fair value at the
end of a long cycle, the deviation in the dollar's value from
its estimated PPP value has often given rise to serious
imbalances on either the internal or external-balance fronts
in the U.S., Europe, or Japan. When those imbalances
reached a critical level, it often set market forces in motion
to bring the overshoot to an end, and to plant the seeds
for a 180-degree shift in the dollar's long-term trend.
Long-Term Cycles in Exchange Rates
Source: Datastream
Source: Datastream
Source: Datastream
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
54 Deutsche Bank Foreign Exchange Research
Structural changes on a country's internal or external-bal-
ance front are largely responsible for the long-term cycles
in exchange rates that we have witnessed. For instance,
the U.S. investment boom of the second half of the 1990s
and the structural rise in U.S. productivity during that pe-
riod are largely responsible for the dollar's persistent rise
over the 1995-2002 period. As illustrated in the first dia-
gram below, the U.S. savings-investment schedule, S-I,
gradually shifted upward and to the left, giving rise to a
persistent uptrend in the dollar's real long-run equilibrium
value over time. Likewise, the persistent rise in Japan's
current-account surplus in the 1980s and early 1990s played
a key role in lifting the yen's equilibrium value over time,
as illustrated in the second diagram.
In both the dollar's and the yen's case, the upward revi-
sions in the real long-run equilibrium exchange rates ap-
pear to have taken place gradually, and not instantaneously
as several theories would have suggested. This would ex-
plain why the up and down cycles in exchange rates en-
dure for years at a time.
Factors Contributing To Long-Term Cycles
The Positive Effects over Time of Structural Increases in Investment
on a Currency’s Long-Run Equilibrium Value
Savings minus Investment,
Current-Account Balance
Real Exchange Rate
Persistent
Structural
Increases in
Investment
Spending
S-I
2
CAB
1
Upward Revision in
the Currency’s Real
Long-Run Equilibrium
Value over Time
q
2
q
1
(0)(-) (+)
S-I
1
S-I
3
q
3
Time
Real Exchange Rate
q
2
q
1
q
3
t
1
t
2
t
3
The Positive Effects over Time of Structural Increases in a Current-Account Surplus
on a Currency’s Long-Run Equilibrium Value
Savings minus Investment,
Current-Account Balance
Real Exchange Rate
Persistent
Structural
Increases in
Current-Account
Balance
CAB
2
CAB
1
Upward Revision in
the Currency’s Real
Long-Run Equilibrium
Value over Time
q
2
q
1
(0)(-) (+)
S-I
1
CAB
3
q
3
Time
Real Exchange Rate
q
2
q
1
q
3
t
1
t
2
t
3
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 55
Most long-term exchange-rate cycles are not driven by a
single shock to the internal or external-balance schedules,
but rather by a series of separate, yet related shocks that
reinforce one another in driving an exchange rate higher or
lower over time. For instance, the dollar's long-term uptrend
between April 1995 and November 2000 reflected a com-
bination of favorable factors including: (1) "New Economy"
forces that helped propel U.S. investment spending and
productivity growth higher, (2) a structural shift in capital
flows from emerging markets and Euroland to the U.S., (3)
higher oil prices, and (4) relatively tight U.S. monetary and
credit conditions up until 2001.
In the case of the euro, its long-term decline since the
early 1990s also reflected a multitude of problems includ-
ing: (1) pervasive structural rigidities in European labor and
product markets, (2) an adverse fiscal/monetary policy mix
in Euroland, (3) higher oil prices, (4) an adverse structural
shift in Euroland capital flows, and (5) excessive pessimism
regarding the potential problems associated with the launch
of the new euro notes and coins in early 2002.
U.S. Dollar Major Currency Index
(1995-2002)
70
80
90
100
110
120
1995 1996 1997 1998 1999 2000 2001 2002
(Index)
U.S. Dollar/Euro Exchange Rate
(and the Euro’s Downtrend since September 1992)
0.80
1.00
1.20
1.40
1.60
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
US$/Euro
Long-Term Cycles in Exchange Rates: The Dollar and Euro in the 1990s
Source: Datastream Source: Datastream
The Dollar’s Long-Term Uptrend
A Stylized Diagram of the Fundamental Forces that Have Contributed to
Increases in the Dollar’s Real Long-Run Equilibrium Value
A
B
C
E
US$ Real
Exchange Rate
Correction
of Initial
Undervaluation
U.S. Investment Boom
Time
q
1
PPP=q
0
F
q
2
Higher Oil Prices
Structural Shift in Portfolio Flows
Tighter Fed Policy
q
4
q
3
D
The Euro’s Secular Decline
A Stylized Diagram of the Fundamental Forces
That Have Contributed to the Euro’s Long-Term Slide
A
B
C
D
The Euro’s
Value
Purchasing Power Parity
Euroland’s Adverse Policy Mix
Time
q
D
E
q
C
Higher Oil Prices
Euroland’s Structural Rigidities
Adverse Structural Shift
in Capital Flows
q
B
q
E
q
A
F
Euro’s Confidence Crisis/
Excessive Pessimism
q
F
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
56 Deutsche Bank Foreign Exchange Research
U.S. Nonresidential Fixed Investment and
Equipment & Software Investment
(Year-over-Year Percentage Changes)
-15
-10
-5
0
5
10
15
20
88 89 90 91 92 93 94 95 96 97 98 99 00 01 02
Equipment & Software----
(%) Fixed Investment____
Source: Datastream
The dollar's rising trend versus most major currencies over
the 1995-2002 period had, for the most part, strong funda-
mental underpinnings. The U.S. economy underwent a
major investment boom over this period, particularly when
compared with the weak investment performance in Eu-
rope and Japan. A large portion of the U.S. boom was con-
centrated in the information technology (IT) arena where
U.S. industry held a dominant, global position. The IT-led
investment boom, in turn, contributed to a surge in U.S.
productivity growth relative to productivity growth else-
where. Since exchange-rate trends are often strongly in-
fluenced by relative productivity trends, the relative surge
in U.S. productivity growth had the effect of raising the
dollar's equilibrium value.
In addition to the rise in productive investment that oc-
curred during the 1995-2000 period, there was probably
also a considerable amount of excess investment during
the height of the IT bubble period, particularly in the
Internet, telecom, and PC areas. This excess investment
spending, which is shown in the diagram below as left-
ward shifts in the savings-investment schedule to S-I3, is
now in the process of being unwound. Indeed, the S-I curve
now appears to be shifting back down to perhaps S-I2, sug-
gesting that the dollar's equilibrium value should be shift-
ing downward as well. At the time of this writing, how-
ever, the dollar's actual value does not appear to have fallen
nearly as much as the apparent decline in its equilibrium
value.
Real Gross Non-Residential Fixed Capital
Formation in the U.S., Euroland, and Japan
(1992-2000)
-40
-20
0
20
40
60
80
100
120
140
1992 1993 1994 1995 1996 1997 1998 1999 2000
Europe---- Japan....
(Cumulative % Change) U.S.____
The 1990s U.S. Investment Boom and the Dollar
Increase in
Capital Inflow
(equal to the gap
between
U.S. domestic
investment
and savings)
The Increase in Investment Spending and the Dollar
Increase in
U.S.
Investment
Spending
Real
Appreciation
of the
U.S. Dollar
Increase in
Future U.S.
Productivity
Growth
The Increase in U.S. Investment Spending
Has Given Rise to an Upward Revision
in the Dollar’s Long-Run Equilibrium Value
Savings minus Investment,
Current-Account Balance
Real Exchange Rate
Structural
Increase in
Investment
Spending
S-I
2
CAB
1
Upward Revision in the
Currency’s Real Long-Run
Equilibrium Value
q
2
q
1
(0)(-) (+)
S-I
1
Excess
Investment
Spending
S-I
3
q
3
“New Economy” Phenomenon
or
Excessive Investment?
Source: Datastream
Source: Datastream
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 57
The Dollars Dramatic Rise in 1999-2000 May Have Been Driven by a Sudden
Upward Revision in the Dollars Real Long-Run Equilibrium Level
Source: Datastream
The dollar’s dramatic surge between the fall of 1999 and
the fall of 2000 may have been propelled by an upward revi-
sion in the market's expectations regarding the U.S.
economy's long-run growth prospects and a subsequent
upward revision in the dollar's real long-run equilibrium value.
The U.S. economy had been growing faster than the
Euroland economy for much of the 1990s, but it wasn't
until late in the decade that observers began to take no-
tice that "something special" was taking place that was
distancing the U.S. from the rest of the world. As the ac-
companying charts show, in annual surveys from 1992-99,
professional economists had been projecting a relatively
bland long-term U.S. economic outlook, with real GDP
growth expected to average 2.5% per annum and produc-
tivity expected to rise by a modest 1.5% per annum over
each ensuing 10-year period.
Normally, one would think that shifts in expectations re-
garding the U.S. long-term growth outlook would take place
gradually over a number of years. But in 1999, expecta-
tions seemed to change overnight. In the 2000 and 2001
surveys, professional economists’ long-term projections
were pushed up sharply. Economists were now forecast-
ing long-term U.S. real GDP growth of 3.3% and U.S. pro-
ductivity growth of 2.5% per annum.
It was as if economists suddenly embraced the notion that
the U.S. was now enjoying a productivity miracle, brought
about by the "New Economy" phenomenon, and that the
gains in productivity would be permanent. Quite likely, a
literature search would show a quantum leap in the num-
ber of academic articles that appeared with the term "New
Economy" in the title beginning in 1999. (For a discussion
of the New Economy's impact on the FX market, see
Liesbeth Van de Craen and Peter Vanden Haute,
The New
Economy and the Foreign Exchange Market
.)
Assuming the marketplace embraced the same long-term
views of the U.S. economy as those held by professional
forecasters, it is not hard to explain why the dollar rose 50
pfennigs versus the Deutschemark between autumn 1999
and autumn 2000. We believe that the suddenly improved
long-term growth outlook for the U.S. economy compelled
market participants to revise up sharply their notion of the
dollar's real long-run equilibrium value. Indeed, the dollar's
rise from the autumn of 1999 to the autumn of 2000, which
followed the upward revision in the U.S. long-term growth
outlook, roughly matched, in terms of magnitude, the rise
that occurred in the prior 4½ years.
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
58 Deutsche Bank Foreign Exchange Research
The "synthetic" euro's real trade-weighted value had been
declining on a trend basis for much of the decade heading
into its 1999 launch date. Yet many pundits expected that
the new single currency would be strong, believing that
the euro would embrace all of the hard-currency attributes
of the Deutschemark and would discard all the soft-cur-
rency attributes of its depreciation-prone ERM members.
This did not happen. By 2001, the euro was trading close
to its 20-year low versus the yen and British pound, and—
except for the 1985 dollar bubble—the euro was trading
close to a record low versus the U.S. dollar.
Perhaps those pundits failed to recognize that the euro's
long-term behavior, being a weighted average of all EMU
member currencies, resembled that of a soft and not a
hard currency. In fact, its persistent decline versus a wide
range of currencies suggests that negative fundamental
developments intrinsic to Euroland probably played a large
role in the euro's long-term slide.
US$/Euro Exchange Rate
(1980-2002)
0.60
0.80
1.00
1.20
1.40
1.60
1.80
1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002
US$/Euro
Synthetic Euro Real Trade-Weighted Index
(1980-2002)
70
80
90
100
110
120
1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002
Bank of England Index
Average
Canadian Dollar/Euro and
Australian Dollar/Euro Exchange Rates
(1980-2002)
0.80
1.00
1.20
1.40
1.60
1.80
2.00
2.20
1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002
A$/Euro----
C
$/Euro____
The Euro Relative to the Euroland Currencies
(European Currencies versus the U.S. Dollar)
20
40
60
80
100
120
140
160
180
75 77 79 81 83 85 87 89 91 93 95 97 99 01
France.... Italy/Spain__ __
(Index) Euro____ DM----
British Pound/Euro Exchange Rate
(1970-2002)
0.50
0.60
0.70
0.80
0.90
1.00
70 72 74 76 78 80 82 84 86 88 90 92 94 96 98 00 02
£/Euro
Japanese Yen/Euro Exchange Rate
(1970-2002)
0
100
200
300
400
500
600
70 72 74 76 78 80 82 84 86 88 90 92 94 96 98 00 02
¥
/Euro
The Euro's Long-Term Struggle
Source: Datastream
Source: Datastream Source: Datastream
Source: Datastream
Source: DatastreamSource: Datastream
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 59
Structural rigidities in Europe's labor and product markets
undoubtedly have adversely affected Europe's long-term
growth potential and thus probably contributed to a down-
ward revision in the market's assessment of the euro's
real long-run equilibrium value as well. Structural rigidities
in Europe's labor market include: job protection legisla-
tion, working time regulations, overly generous early re-
tirement schemes, unemployment insurance and welfare
benefits, and barriers to hiring skilled labor from abroad.
On the product front, structural rigidities include the slug-
gishness of firms to adopt a culture of risk taking, the high
cost of telecommunications, and the unwillingness of many
European governments to allow market forces to operate
unfettered. The charts below show that European labor
markets are far more regulated than the U.S., European
unemployment benefits are far more generous, while Eu-
ropean taxes and real labor costs are higher than they are
in the U.S. as well.
Strictness of Labor Market Regulation
(Most to Least Strict)
0
2
4
6
8
G
reece
Italy
Spain
France
Sweden
EMU-11
Germany
Ireland
Finland
Portugal
Norway
Austria
Belgium
Netherlands
Switz.
Denmark
U.K.
U.S.
A
ggregate Index
Source: OECD
Overall Taxes as a Percentage of Earnings
30
35
40
45
50
55
60
65
D
enmark
Belgium
Sweden
France
Germany
Norway Italy
Finland
Ireland
Netherlands
EMU-11
Portugal
Spain U.K. U.S.
(%)
Source: OECD
Unemployment Benefit Generosity
(Most to Least Generous)
16
19
37
49
50.1
51
54
55
5959
6262
67
69
81
0
20
40
60
80
100
Denmark
Netherlands
Sweden
Norway
Switz.
Belgium
Finland
France
Germany U.K.
EMU-11
Spain
Ireland Italy U.S.
A
ggregate Index
Source: OECD
Europe’s Real Labor Costs & Total Employment
80
100
120
140
160
180
1970 1975 1980 1985 1990 1995 1998
Total Employment----
(1970 = 100) Real Labor Costs____
Source: OECD
2001 Global Competitiveness
(based on World Economic Forum Rankings)
0
20
40
60
80
Finland
U.S.
Canada
Singapore
Australia
Norway
Taiwan
Netherlands
Sweden
N.Z.
Ireland
U.K.
Hong Kong
Denmark
Switz.
Iceland
Germany
Austria
Belgium
France
Japan
Spain
Korea
Israel
Portugal
Italy
Chile
Hungary
Estonia
Malaysia
Slovenia
Mauritius
Thailand
South Africa
Costa Rica
Greece
Rankings(Most to Least Competitive)
Source: World Economic Forum
(Top-36 Countries)
European and U.S. Total Employment
95
100
105
110
115
120
125
130
135
84 86 88 90 92 94 96 98 00 02e
U.S.----
(1984=100) Europe____
Source: OECD Economic Outlook
Euroland's Structural Rigidities and the Euro's Real Long-Run
Equilibrium Value
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
60 Deutsche Bank Foreign Exchange Research
Since the beginning of floating exchange rates, there have
been a number of occasions when exchange rates have
significantly overshot their long-run equilibrium path, ei-
ther on the upside or the downside. We define an exchange-
rate overshoot episode as one in which an exchange rate
deviates by more than +/- 20% from its long-run purchas-
ing power parity path and/or where there exists a signifi-
cant divergence between the actual trend in exchange rates
and some long-term, key determinant—such as the trend
in real interest-rate differentials in the case of the Deut-
schemark, or the trend in relative monetary-base growth
rates in the case of the Japanese yen.
For the most part, long-term cycles have strong fundamen-
tal underpinnings. That is, we can find a strong positive
relationship between the long-term trend in exchange rates
and the long-term trend in some key fundamental indica-
tor. What we find interesting, however, is that at the end
of a typical long-term dollar cycle there has been a ten-
dency for the dollar's value to overshoot its long-run fun-
damentally driven equilibrium path by rather large amounts.
A simple model can be designed to explain why exchange
rates tend to overshoot. The model assumes that inves-
tors' expectations of exchange rates are driven by both
fundamental and technical considerations. Mathematically,
we can say that exchange rate expectations, eet, in time
period
t
are driven by fundamental forces, fundt, and by
the past pattern of exchange-rate movements, et-1.
It is further assumed that the market will assign weights
to the relative importance that it attaches to the current
trend in fundamental forces,
w
, and to the past pattern of
exchange-rate movements, 1-
w
. Those weights are as-
sumed to vary over time, with the market assigning more
weight to fundamentals and less weight to technicals, and
vice versa, as fundamental and technical forces vary in
terms of influencing the path that currencies take. Ex-
change-rate expectations can therefore be expressed as:
eet =
w
(fundt) + (1-
w
)et-1
A Stylized Model of Exchange-Rate Overshooting
at the End of a Long-Term Uptrend
Currency’s
Value
Time
Economic Fundamentals
Exchange Rate Expectations can be expressed as:
e
et = w (fundt) + (1-w )et-1
where:
eet= exchange-rate expectations
fundt= fundamental forces
et-1 = past pattern of exchange-rate movements
w = market-assigned weight to the importance of the current
trend in fundamental forces
1-w = market-assigned weight to the importance of the past pattern
of exchange-rate movements
Overshooting Exchange Rates
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 61
Typically, in the early and middle stages of an extended
exchange-rate cycle, the market tends to attach greater
weight to the trend in economic fundamentals (i.e.,
w
rises). This is because the trend in exchange rates and the
trend in key fundamental variables parallel each other quite
closely. In the latter stages of an extended exchange-rate
cycle, however, there often occurs some disconnect be-
tween the trend in the exchange rate and the trend in key
fundamental variables.
What causes this disconnect to occur is the tendency of
the market at the end of a long cycle to assign less weight
to fundamental forces and more weight to the prevailing
trend in exchange rates (i.e., 1-
w
rises). For instance, after
a long cycle of currency appreciation, the market gets ac-
customed to being long the appreciating currency and
tends to shrug off any adverse underlying fundamental
developments as being just a temporary phenomenon.
Hence, the market will begin to attach less weight to the
deteriorating trend in fundamentals (i.e.,
w
will gradually
fall) and more weight to the past favorable trend in the
exchange rate (i.e., 1-
w
will gradually rise).
By attaching greater weight to expectations that the pre-
vailing trend will remain strong and less weight to the de-
terioration of fundamental forces, the trend in exchange
rates and the trend in fundamentals becomes discon-
nected. Eventually, a point will be reached when the mar-
ket finally recognizes that the deterioration in fundamen-
tals is a permanent, and not just a temporary phenom-
enon. At that point, market expectations will shift, result-
ing in greater weight being assigned to deteriorating fun-
damental forces and less weight to the prospect that the
exchange-rate overshoot will persist. It is at this stage that
the unwinding of long speculative positions gives rise to a
rapid, and at times violent, correction of the exchange-rate
overshoot.
There have been several extraordinary episodes of major
currency overshooting since the beginning of floating ex-
change rates—the Deutschemark in the mid-1980s, the
yen in 1995 and 1998, and the dollar in general in 2000-02.
In the case of the Deutschemark, there was a significant
disconnect between the trend in U.S./German real long-
term interest-rate differentials and the DM/US$ exchange
rate towards the end of the dollar's remarkable run up-
ward during its bubble period. There was simply no way to
explain the dollar's surge in 1984-85 on the basis of real
yield spreads, and this overshoot in the dollar's value was
eventually corrected in 1985-87.
In the case of the yen, there was a similar disconnect be-
tween the trend in U.S. and Japanese relative monetary-
base growth and the trend in the yen's value versus the
dollar in 1995 and again in 1998. The dollar overshot to the
downside in 1995 and overshot to the upside in 1998, and
in both cases the overshoots were subsequently corrected.
Source: Datastream
Source: Datastream
The Market Attaches Less Weight
to Fundamental Forces During
an Exchange-Rate Overshooting Period
Market-Assigned
Weight
(w)
Time
Market-Assigned
Weight
(1-w)
Beginning of
Exchange-Rate Cycle Overshooting
Period
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
62 Deutsche Bank Foreign Exchange Research
It is widely recognized that financial markets need stabiliz-
ing speculators to ensure that financial market prices do
not wander too far from fair value. By buying at price lev-
els that are perceived to be low and selling at price levels
that are perceived to be high, speculators will ensure that
asset prices do not become too stretched relative to fair
value. If there is a fairly elastic supply of speculative capi-
tal in the foreign-exchange market, a significant increase
in commercial demand for dollars (perhaps generated by a
foreign acquisition of a U.S. firm) need not give rise to a
dramatic rise in the dollar's value. Instead, the increased
commercial demand for dollars may give rise to only a
modest increase in the dollar's value from e0 to e1 as shown
in the diagram below.
However, consider a situation where there is an absence
or a shortage of stabilizing speculative capital. This might
occur if hedge funds, investors, and foreign-exchange trad-
ers are unwilling or unable to commit relatively large sums
of capital for speculative purposes. In such a case, the
supply of speculative capital in the foreign-exchange mar-
ket would appear to be relatively inelastic. In that environ-
ment, it may only take a modest increase in the commer-
cial demand for dollars to push the dollar's value sharply
higher from e0 to e2. In other words, when there is a short-
age of stabilizing speculative capital, exchange rates are
likely to be more volatile and prone to overshooting, even
amid modest changes in the commercial demand for dol-
lars.
Such behavior might explain why the dollar has remained
significantly overvalued against a wide range of curren-
cies in the past few years. With many investors, hedge
funds, and foreign-exchange traders unwilling or unable to
risk sufficient capital to bet against the dollar, there simply
has not been enough speculative capital in the market to
reverse the dollar's overshoot.
The tendency of speculative capital to dry up when mar-
ket prices move far out of line with fair value is not unique
to the foreign-exchange market. Indeed, most speculative
bubbles, particularly at their peaks, are not fed by exces-
sive speculation, but instead are driven by an absence of
stabilizing speculation. When an absence of stabilizing
speculative capital causes exchange rates to overshoot
their long-run equilibrium value, central banks often inter-
vene and take on the role of a stabilizing speculator.
How a Shortage of Stabilizing Speculative Capital Can Give Rise to
Exchange-Rate Overshooting
An Absence of Currency Speculation
Can Lead to Greater Exchange-Rate Volatility
and Exchange-Rate Overshooting
Quantity of Dollars
Exchange Rate
C
Commercial Demand for Dollars
2
Commercial Demand for Dollars
1
Inelastic Speculative
Supply of Dollars
B
A
Elastic Speculative
Supply of Dollars
e
2
e
1
e
0
Q
C
Q
B
Q
A
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 63
Exchange-Rate Determination
in the Medium Term
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
64 Deutsche Bank Foreign Exchange Research
Exchange rates are simultaneously determined by long-
term structural, medium-term cyclical, and short-term
speculative forces. Since long-term structural forces often
change at a glacial pace, they are likely to exert a greater
influence on a currency's long-run equilibrium path than
on a currency's short or medium-term path. A variety of
structural forces operate jointly to influence a currency's
long-run equilibrium path. These include the long-term trend
in national inflation rates, relative productivity growth, long-
term trends in net foreign asset and liability positions, per-
sistent trends in a country's terms of trade, and long-term
trends in national savings and investment.
Medium-term cyclical forces often cause a currency to ei-
ther rise or fall relative to its long-run equilibrium path. In
many cases the medium-term deviations from the long-
run equilibrium path can be quite sizeable, and in some
instances they can be quite persistent. In the long run,
these medium-term cyclical forces tend to wash out, and
once they do, there is a tendency for the exchange rate to
return to its long-run equilibrium path.
Nevertheless, in the interim, fund managers must find a
way to adjust their portfolios to account for these devia-
tions from long-run equilibrium levels or else risk serious
underperformance over medium-term horizons. Since in-
vestment performance is often evaluated over relatively
short time spans today, it is important for fund managers
to channel their energies towards getting the direction that
exchange rates take over short and medium-term horizons
right.
Medium-term trends in exchange rates tend to be influ-
enced by the direction of macroeconomic variables. The
key medium-term determinants of exchange rates include
trends in real interest-rate differentials, current and capital
account trends, relative monetary and fiscal policies, rela-
tive economic growth, and portfolio-balance considerations.
The success that these variables have had in explaining
the medium-term trends that exchange rates have taken
in the past is at best mixed. In many cases, one can find
evidence of a strong correlation between one or more of
these variables and the trend in exchange rates over sev-
eral cycles. But something often happens to break this
tight cyclical linkage. When it does, the subsequent use-
fulness of that variable as an explanatory variable becomes
seriously undermined.
Overall, we find that shifts in fiscal and monetary policies
play the most important role in driving a currency's value
on a medium-term basis, with many of the major exchange-
rate cycles in the 1990s attributable to shifts in fiscal/mon-
etary policy mixes of individual countries. We find that the
trend in the ratio of Japan's monetary base to the U.S.
monetary base has done the best job of explaining the
Japanese yen/U.S. dollar exchange rate's medium-term cy-
clical path over the past 12 years. In the case of the Deut-
schemark (and now the euro), the trend in the U.S./Ger-
man real long-term yield spread has done the best job of
explaining the medium-term cyclical paths that the Deut-
schemark/U.S. dollar exchange rate has taken over the past
25 years.
The Yen and Japanese/U.S. Monetary Policies
0.90
1.00
1.10
1.20
1.30
1.40
1.50
80
100
120
140
160
180
89 90 91 92 93 94 95 96 97 98 99 00 01 02
Japan/U.S. Monetary Base(ratio)----
Y
en/US$____
Note: Y2K Adjusted,
Japan Reserve Require. Rate Change Adj. Series
Exchange-Rate Determination in the Medium Term
DM/US$ Exchange Rate and U.S./German
Real Interest-Rate Differentials
(10-Year Bond Yields Less CPI)
-1000
-800
-600
-400
-200
0
200
400
600
1.00
1.50
2.00
2.50
3.00
3.50
75 77 79 81 83 85 87 89 91 93 95 97 99 01
U.S. Less German Real Interest Rates(bp)----DM/US$____
Source: Datastream
Source: Datastream
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 65
International parity conditions are financial arbitrage con-
ditions that would exist in an ideal world. The key interna-
tional parity conditions are (1) purchasing power parity, (2)
covered interest-rate parity, (3) uncovered interest-rate
parity, (4) the Fisher effect, which focuses on inflation as
the key determinant of nominal interest rates, and (5) the
forward exchange rate as an unbiased predictor of the fu-
ture spot rate. Parity conditions show how inflation differ-
entials, interest-rate differentials, forward exchange rates,
and expected changes in exchange rates are linked inter-
nationally. Parity conditions tell us that high (low) inflation
countries should see their currencies depreciate (appreci-
ate) over time, and that forward exchange rates should
function as unbiased predictors of future spot rates. These
international parity conditions form the building blocks of
many of the major macro models of exchange-rate deter-
mination.
Empirical evidence indicates that most of the key interna-
tional parity conditions rarely hold in either the short or
medium term. There are often significant departures from
purchasing power parity, interest-rate differentials often
fail to explain future exchange-rate changes, and the for-
ward exchange rate has been found to be a poor predic-
tor of the future spot exchange rate. Why study interna-
tional parity conditions if they fail to hold? Actually, if inter-
national parity conditions held at all times, there would
be no profitable arbitrage opportunities available for inter-
national investors to exploit by moving capital from one
market to another. It is only when international parity con-
ditions fail to hold that profit opportunities from cross-
border investments become available.
Ex Ante
Purchasing Power Parity
According to
ex ante
purchasing power parity (PPP), the expected
change in the spot exchange rate should equal the difference in
expected national inflation rates.
Ex ante
PPP tells us that a high-
inflation country should see its currency depreciate and that a
low-inflation country should see its currency appreciate over time.
Covered Interest-Rate Parity
According to the covered interest-rate parity condition, an invest-
ment in a foreign-currency deposit completely hedged against
exchange-rate risk should yield exactly the same return as a com-
parable domestic-currency deposit. Since a hedged foreign-cur-
rency investment will have the same risk characteristics as a do-
mestic-currency investment, we should expect the yield on the
domestic investment, iD, to equal the foreign interest rate, iF, less
the forward discount, FD. Indeed, arbitrage should insure that
this would always be the case. The empirical evidence in support
of covered interest-rate parity is quite robust.
Uncovered Interest-Rate Parity
According to the uncovered interest-rate parity (UIP) condition,
the expected return on an uncovered foreign-currency investment
should equal the expected return on a comparable domestic-cur-
rency investment. The expected return on a domestic-currency
investment, iD, is known with certainty, while the expected return
on an uncovered foreign-currency investment, iF- ee, is not known
with certainty because the actual change in the exchange rate, e,
may turn out to be different from the expected change in the
exchange rate, ee. The UIP condition implies that investors do not
need to be compensated—in the form of a risk premium—for
taking on the risk that their expectations may prove to be wrong.
In the absence of a risk premium, the yield spread between for-
eign and domestic currency deposits, iF - iD, should adjust to equal
the expected change in the exchange rate, ee. The empirical evi-
dence is not highly supportive of UIP in either the short or me-
dium run.
e = p - p
eee
FD
Expected Change
in Exchange Rate
Foreign-Domestic
Expected
Inflation
Differential
=
i - i = FD
FD
Foreign-Domestic
Interest-Rate
Differential
Forward
Discount
=
e = i - i
e
FD
Foreign-Domestic
Interest-Rate
Differential
Expected
Change in
Exchange Rate
=
International Parity Conditions—
The Building Blocks of Macro Exchange-Rate Modeling
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
66 Deutsche Bank Foreign Exchange Research
Fisher Effect
According to the Fisher effect, the nominal interest rate, i,
in a given country will equal the real interest rate, r, plus
the expected inflation rate, pe. If the real interest rate in
the foreign country is equal to the domestic real interest
rate, rF = rD, then the yield spread between two countries,
iF - iD, should equal the expected inflation differential be-
tween the two countries, pFe - pDe. Empirical evidence sug-
gests that real interest rates often differ among countries.
Thus, nominal yield spreads will not necessarily reflect dif-
ferences in national inflation rates.
Forward Rate as an Unbiased Predictor of the
Future Spot Rate
If covered interest-rate parity holds such that iF- iD=FD, and
uncovered interest-rate parity holds such that ee= iF- iD, then
the forward discount, FD, will equal the expected change
in the spot exchange rate, ee. Empirical evidence suggests,
however, that the forward exchange rate is a poor and bi-
ased predictor of the future spot exchange rate.
International Parity Conditions—
How Spot Exchange Rates, Forward Exchange Rates,
and Interest Rates Are Linked Internationally
If all of the key international parity conditions held at all
times, the expected change in the spot exchange rate
would equal (1) the forward discount, (2) the interest-rate
differential, and (3) the expected inflation differential. These
conditions will hold only in an ideal world. Empirical stud-
ies actually reject most of the parity conditions (covered
interest-rate parity being the exception). There are often
large and persistent deviations from PPP and UIP, and the
forward exchange rate has been found to be both a poor
and a biased predictor of the future spot rate. When these
parity conditions fail to hold, profitable investment oppor-
tunities become available.
e = FD
e
Forward
Discount
Expected
Change in
Exchange Rate
=
i - i = p - p
FD F D
ee
Foreign-Domestic
Interest-Rate
Differential
=
Foreign-Domestic
Expected
Inflation
Differential
Expected
Change in
Exchange Rate
e
e
Foreign-Domestic
Expected
Inflation
Differential
p - p
FD
ee
Foreign-Domestic
Interest-Rate
Differential
i - i
FD
Forward
Discount
FD
e = p - p
eee
FD
i - i = FD
FD
e = i - i
e
FD
i - i = p - p
FD F D
ee
e = FD
e
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 67
Derivation of the Real Interest-Rate Differential Model
of Exchange-Rate Determination
(1) ee$ = iF - iUS Uncovered Interest-Rate Parity
(2) ee$ = qe$ - (peF - peUS)Ex Ante Purchasing Power Parity
(3) qe$ = (iF - peF ) - (iUS - peUS)(2) minus (1)
(4) qe$ = rF - rUS Real Interest-Rate Parity
(5) qe$ = q$ - q$Long-Run Exchange-Rate Adjustment
(6) q$ - q$ = rF - rUS Combine (4) and (5)
(7) q$ = q$ + (rUS - rF)RID Model
(8) q$ = q$ + (rUS - rF) - US - φF)Extended RID Model
International capital flows have become increasingly mo-
bile in the past 20 years as barriers to cross-border move-
ments of capital have been gradually lifted in most mar-
kets. As cross-border capital flows have risen in importance,
interest-rate differentials have come to play a more impor-
tant role in the determination of exchange rates. Indeed,
empirical evidence has shown that relative current-account
trends have become less important in explaining exchange-
rate movements in the past two decades, while interest-
rate differentials have become more important.
To analyze the role that interest-rate differentials play in
the determination of exchange rates, many international
economists use the real interest-rate differential (RID)
model as their principal analytical tool. The RID model is a
formal model that links the trend in real exchange rates to
the trend in real interest-rate differentials.
The RID model is fairly simple to derive. If it is assumed
that both uncovered interest-rate parity (equation 1) and
ex ante purchasing power parity (equation 2) hold in the
long run, then the expected long-run change in the dollar's
real value will equal the real long-term yield spread (equa-
tions 3 and 4). If it is assumed that the expected change in
the dollars real value in the long run equals the difference
between the dollar’s actual real value and its long-run equi-
librium level (equation 5), it can then be shown in equation
7 that the dollar's actual real value, q$, can be expressed
as a function of two key variables—the dollar's real long-
run equilibrium value, q$, and the real long-term interest-
rate differential, rUS-rF. According to this model, the dollar's
real value would rise if there were an upward revision in
the market's assessment of the dollar's real long-run equi-
librium value and/or if there were a rise in the U.S./foreign
real long-term interest-rate differential.
The RID model can be extended to incorporate shifts in
risk premiums on U.S., φUS, and foreign assets, φF. The ex-
tended model is shown in equation 8, where a rise in the
risk premium on U.S. assets tends to exert downward pres-
sure on the dollars value, while a rise in the risk premium
on foreign assets tends to exert upward pressure on the
dollars value.
The Real Interest-Rate Differential Model
of Exchange-Rate Determination
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
68 Deutsche Bank Foreign Exchange Research
Empirical studies have found that the trend in real long-
term yield spreads does a better job of explaining exchange-
rate movements than the trend in nominal short-term yield
spreads. There are two reasons for this. First, real yields—
rather than nominal yields—better reflect structural shifts
in savings and investment and the easing/tightening moves
by central banks. Second, changes in long-term interest
rates are likely to have a more profound impact on exchange
rates than changes in short-term rates because changes
in short-term rates can be and often are transitory, while
changes in long-term rates tend to be sustained. Taking
both of these factors together, the historical evidence sup-
ports a stronger linkage between exchange rates and real
long-term yield spreads than between exchange rates and
nominal short-term yield spreads.
The accompanying diagram shows that the magnitude of
an exchange-rate response to a given change in real inter-
est-rate spreads is far greater when real long-term yield
spreads change (from q0 to q5) than when real short-term
yield spreads change (from q0 to q1). That is because ex-
change rates tend to respond more vigorously to changes
in real interest-rate spreads that are expected to persist.
The diagram also suggests that large changes in exchange
rates may result from fairly modest changes in real long-
term interest-rate differentials. The diagram illustrates as
well that in the long run, the response of an exchange rate
to a change in real yield spreads will only be transitory.
In theory, the rise in the real exchange rate in response to
the change in real yield spreads should be instantaneous,
followed by a gradual decline in the real exchange rate back
to its long-run equilibrium value as real yield spreads re-
turn to their normal levels. In practice, both the change in
real yield spreads and the change in exchange rates tend
to be much more gradual than the diagram depicts. It is
this more gradual response that gives rise to the tight
graphical relationship that one finds when comparing the
trend in the dollar's value with the trend in U.S./foreign
real long-term yield spreads.
Real Exchange-Rate Response to a Rise in Real
Short- and Long-Term Interest-Rate Differentials
U.S. Dollar’s Real Value
Time
q
2
Source: Adapted from Clark . (1994).
et al
q
1
t
0
t
1
t
2
t
3
t
4
t
5
q= q
0
q
3
q
4
q
5
An increase in real
interest-rate differentials
expected to last for
five years.
The Real Interest-Rate Differential Model—Theory
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 69
The trend in real long-term interest-rate differentials has
played a pivotal role in driving the dollar's value versus the
Deutschemark in the past few decades (as shown below
in the charts of the various periods).
The decline of the dollar in the late 1970s, the rise of the
dollar in the first half of the 1980s, and the dollar's decline
after 1985 can be attributed to changes in U.S./German
real long-term yield spreads.
Over the 1993-98 period, an adverse shift in real yield
spreads led to the dollar's decline in 1994-95, while the
dollar's rise in 1995-98 can be attributed to a shift in the
U.S./German real yield spread that was favorable to the
dollar.
In the 1999-2000 period, the relationship broke down when
the trend in Deutschemark (now the euro) completely di-
verged from the trend in the U.S./German (now the U.S./
European) real long-term yield spread. Although real yield
spreads moved against the U.S., the dollar nevertheless
surged against the euro, with the dollar’s gains likely at-
tributable to a rise in the dollars real long-run equilibrium
level.
We believe that the breakdown in the real yield spread/
exchange rate linkage in 1999-2000 will prove to be only a
temporary phenomenon. The indications are now that the
real-interest-rate-spread/exchange-rate linkage has been re-
established in 2001-02.
The Real Interest-Rate Differential Model—Evidence
DM/US$ Exchange Rate and U.S./German
Real Interest-Rate Differentials
(10-Year Bond Yields Less CPI, 1977-1990)
-800
-600
-400
-200
0
200
400
600
1.00
1.50
2.00
2.50
3.00
3.50
77 78 79 80 81 82 83 84 85 86 87 88 89 90
U.S. Less German Real Interest Rates(bp)----DM/US$____
DM/US$ Exchange Rate and U.S./German
Real Interest-Rate Differentials
(10-Year Bond Yields Less CPI, 1993-June 1999)
-300
-200
-100
0
100
200
1.30
1.40
1.50
1.60
1.70
1.80
1.90
2.00
1993 1994 1995 1996 1997 1998 1999
U.S. Less German Real Interest Rates(bp)----DM/US$____
DM/US$ Exchange Rate and U.S./German
Real Interest-Rate Differentials
(10-Year Bond Yields Less CPI, July 1999-2000)
-120
-100
-80
-60
-40
-20
0
20
1.80
1.90
2.00
2.10
2.20
Jul-99 Oct-99 Jan-00 Apr-00 Jul-00
U.S. Less German Real Interest Rates(bp)----DM/US$____
DM/US$ Exchange Rate and U.S./German
Real Interest-Rate Differentials
(10-Year Bond Yields Less CPI, Aug. 2000-2002)
-150
-100
-50
0
50
100
150
2.05
2.10
2.15
2.20
2.25
2.30
2.35
Aug-00 Nov-00 Feb-01 May-01 Aug-01 Nov-01 Feb-02
U.S. Less German Real Interest Rates(bp)----DM/US$____
Source: Datastream Source: Datastream
Source: Datastream Source: Datastream
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
70 Deutsche Bank Foreign Exchange Research
To many FX market participants, 2001 was a highly un-
usual year. The Federal Reserve cut short-term interest
rates from 6.5% to 1.75% and U.S./foreign short-term yield
spreads narrowed dramatically, yet the dollar remained
remarkably resilient. This has led some observers to ques-
tion whether the old rules for FX forecasting still apply.
Indeed, some observers have tried to redefine the linkage
between interest rates and exchange rates by arguing that
lower interest rates may, in fact, be positive for a currency
if they raise expectations of future economic growth in
both absolute and relative terms.
From an historical vantage point, 2001 does not appear to
be that unusual. Indeed, 2001 is not the first time that
nominal U.S. short-term interest rates and the dollar's value
have moved in opposite directions. Consider the Fed tight-
ening episodes of 1972-74, 1977-80, 1986-88, and 1994-
95. In all of those cases, the dollar actually fell, and in most
cases, quite sharply.
Nominal Short-Term Yields and the Dollars Value
The Dollar & U.S. Short-Term Interest Rates
(January 1972-August 1974)
4
6
8
10
12
14
16
2.20
2.40
2.60
2.80
3.00
3.20
3.40
1972 1973 1974
3-Month Euro-$ Rate(%)----DM/US$____
The Dollar & U.S. Short-Term Interest Rates
(January 1977-March 1980)
0
5
10
15
20
25
1.60
1.80
2.00
2.20
2.40
2.60
1977 1978 1979 1980
3-Month Euro-$ Rate(%)----DM/US$____
The Dollar & U.S. Short-Term Interest Rates
(August 1986-December 1988)
5
6
7
8
9
10
1.60
1.70
1.80
1.90
2.00
2.10
1986 1987 1988
3-Month Euro-$ Rate(%)----DM/US$____
The Dollar & U.S. Short-Term Interest Rates
(1994-1995)
3
4
5
6
7
1.30
1.40
1.50
1.60
1.70
1.80
1994 1995
3-Month Euro-$ Rate(%)----DM/US$____
Source: Datastream Source: Datastream
Source: Datastream Source: Datastream
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 71
The seemingly anomalous relationship between nominal
short-term interest rates and the dollars value that occurred
in 1972-74, 1977-80, 1986-88, and 1994-95 can be explained
by the trend that real long-term interest-rate differentials
took in all four episodes. In all of those cycles, real long-
term yield spreads moved in favor of Germany, which sup-
ported the case for a weaker, not a stronger, dollar.
In 2001, the same thing happened, except in reverse. Even
though nominal short-term yield spreads moved in Europe's
favor, real long-term yield spreads actually moved mod-
estly in favor of the dollar.
The conclusion one should draw from this historical analy-
sis is that the level and trend in nominal short-term yield
spreads may not be the best guide for explaining medium-
term exchange-rate movements. Rather, it is the level and
trend in real long-term yield spreads that do the best job in
terms of explaining medium-term exchange-rate move-
ments.
Real Long-Term Yield Spreads and the Dollars Value
The Dollar and U.S./German
Real Interest-Rate Differentials
(10-Year Bond Yields Less CPI, 1972-1974)
-1000
-800
-600
-400
-200
0
200
2.20
2.40
2.60
2.80
3.00
3.20
3.40
1972 1973 1974
U.S. Less German Real Interest Rates(bp)----DM/US$____
The Dollar and U.S./German
Real Interest-Rate Differentials
(10-Year Bond Yields Less CPI, 1977-1980)
-800
-600
-400
-200
0
1.60
1.80
2.00
2.20
2.40
2.60
1977 1978 1979 1980
U.S. Less German Real Interest Rates(bp)----DM/US$____
The Dollar and U.S./German
Real Interest-Rate Differentials
(10-Year Bond Yields Less CPI, 1986-1988)
-250
-200
-150
-100
-50
0
50
1.40
1.60
1.80
2.00
2.20
2.40
2.60
1986 1987 1988
U.S. Less German Real Interest Rates(bp)----DM/US$____
The Dollar and U.S./German
Real Interest-Rate Differentials
(10-Year Bond Yields Less CPI, 1994-1995)
-250
-200
-150
-100
-50
0
50
100
150
1.30
1.40
1.50
1.60
1.70
1.80
1994 1995
U.S. Less German Real Interest Rates(bp)----DM/US$____
Source: Datastream Source: Datastream
Source: Datastream Source: Datastream
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
72 Deutsche Bank Foreign Exchange Research
Numerous academic papers have investigated empirically
whether the forward exchange rate has served as a reli-
able unbiased predictor of the future spot exchange rate.
The forward-rate predictor hypothesis is usually tested by
regressing the change in the exchange rate, et+k, on the
forward discount, FDt,k:
et+k = a + ß (FDt,k)
If the forward exchange rate were an unbiased predictor
of the future spot rate, estimates of the regression coeffi-
cient, beta (ß), would be equal to 1.0. If ß = 1.0, then et+k
would be expected to equal FDt,k, on average, over time.
Interestingly, the academic evidence strongly rejects the
hypothesis that the forward exchange rate is an unbiased
predictor of the future spot exchange rate, with estimates
of ß that are often significantly less than one. Indeed, ß
has often been found to be less than zero. In a survey of
75 published papers on the forward rate predictor hypoth-
esis, the average estimate of ß was found to be - 0.88.
This startling finding has the following implication for cur-
rency investment strategies. Currencies that trade at a for-
ward discount, on average, weaken less than the amount
implied by the forward discount. If anything, there has been
a tendency for such currencies to appreciate, not depreci-
ate, over time. This means that the forward discount has
been a biased predictor of future changes in the spot ex-
change rate. Hence, the term "forward discount bias." The
opposite applies for currencies that trade at a forward pre-
mium. From a strategy standpoint, if these findings on past
performance were to remain valid in the future, investors
would stand to benefit by over-weighting currencies that
trade at a forward discount and under-weighting curren-
cies that trade at a forward premium.
Various interpretations have been offered to explain these
anomalous findings. One view holds that currencies that
trade at a forward discount are more risky than those that
trade at a forward premium. Hence, investors demand a
higher risk premium to buy and hold such currencies, and
the excess return earned by investing in such currencies
merely represents the reward for accepting higher risk.
Another view holds that the market simply makes repeated
expectational errors, and that investors could exploit this
by trading against the implicit views expressed in the for-
ward exchange market.
Although excess returns could have been earned in the
past by investors who sought to exploit the "bias" in the
forward discount, the risk/return tradeoff is not always at-
tractive for such strategies. Monthly returns can be and
often are quite volatile for single-currency trades. In fact,
single-currency trading strategies have not been attractive
enough to justify risking large sums to trade the bias. (A
single-currency strategy consists of buying, or going long,
only the highest yielding currency, i.e., the currency with
the largest forward discount, and funding it by borrowing
(or going short) only the lowest yielding currency, i.e., the
currency with the largest forward premium.)
Risk-adjusted excess returns from following diversified
multi-currency approaches to exploit the bias, however,
have been more impressive. Adopting a diversified strat-
egy of going long the three highest-yielding currencies in
the industrial world and funding that position by going short
the three lowest-yielding currencies has yielded estimated
Sharpe ratios averaging 0.7-0.8 for dollar, euro, and yen-
based investors. That is nearly double the Sharpe ratios
generated by buying and holding the shares of the S&P
500 index. This basket approach to trading the forward dis-
count bias is the bedrock of Deutsche Bank's Forward-
Rate Bias trading system.
Forward-Rate Bias
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 73
Our favorite approach to trading the forward-rate bias is to
adopt a diversified strategy to exploit the fact that curren-
cies trading at a forward discount tend to outperform those
currencies trading at a forward premium. The 11 curren-
cies that we include in our trading system are the U.S.
dollar, euro, Japanese yen, Swiss franc, British pound,
Canadian dollar, Australian dollar, New Zealand dollar, Dan-
ish krone, Norwegian krone, and Swedish krona. Our in-
vestment approach is straightforward: we recommend
going long the three highest-yielding currencies in the in-
dustrial world and going short the three lowest-yielding
currencies in the industrial world. Net long/short positions
are put on at the beginning of each month and then closed
at the end of each month. This process is repeated each
month over time.
The table below shows how our monthly recommenda-
tions for long and short currency positions for the 11 in-
dustrial country currencies can vary over time. At the be-
ginning of 2001, a high-yield/low-yield currency investment
strategy would have favored being long the Norwegian
krone, New Zealand dollar, and U.S. dollar, and being short
the Swedish krona, Swiss franc, and Japanese yen. By
the end of 2001, with U.S. yields having fallen below those
in most other markets, this strategy shifted in favor of be-
ing long the Norwegian krone, New Zealand dollar, and
Australian dollar, and short the Swiss franc, U.S. dollar, and
Japanese yen.
Recommended Long and Short Positions versus the U.S. Dollar
from Adhering to Long High-Yielders/Short Low-Yielders Currency-Investment Strategy
(January 2001-April 2002)
Month AUD CAD DKK EUR JPY NZD NOK SEK CHF GBP USD
Jan-01 ----short long long short short - long
Feb-01 ----short long long short short long -
Mar-01----short long long short short long -
Apr-01----short long long short short long -
May-01----short long long short short long -
Jun-01 ----short long long - short long short
Jul-01 ----short long long - short long short
Aug-01 ----short long long - short long short
Sep-01 ----short long long - short long short
Oct-01 ----short long long - short long short
Nov-01 long ---short long long - short - short
Dec-01 long ---short long long - short - short
Jan-02 long ---short long long - short - short
Feb-02 long ---short long long - short - short
Mar-02 long ---short long long - short - short
Apr-02 long ---short long long - short - short
May-02----short long long long short - short
Note: Datastream is the source of the underlying data.
Trading the Bias
Three-Month Nominal Interest Rates
(as of December 31, 2000)
0.6
3.4
4.1
4.9
5.3
5.7
5.8
6.1
6.3
6.5
7.4
0
2
4
6
8
Norway N.Z. U.S. Aust. U.K.Canada
Denmark Euro
Sweden
Switzerland Japan
(%) 3-Month Euro-Deposit Rates
Three-Month Nominal Interest Rates
(as of December 31, 2001)
0.1
1.81.8
2.1
3.3
3.6
3.8
4.0
4.2
4.8
6.5
0
1
2
3
4
5
6
7
Norway N.Z. Aust. U.K.
Sweden
Denmark EuroCanada
Switzerland U.S. Japan
(%) 3-Month Euro-Deposit Rates
Source: Datastream Source: Datastream
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
74 Deutsche Bank Foreign Exchange Research
Our analysis of the risks and returns that would have been
generated for a dollar-based investor following our diversi-
fied forward-rate bias strategy indicates that relatively at-
tractive risk-adjusted returns could have been earned in
the past 16 years. Over the 1986-2001 period, the average
annual excess return from following the forward-rate bias
strategy was around 5.6%, with an annualized standard
deviation of 7.8% and a Sharpe ratio of 0.73. This Sharpe
ratio compares favorably with the annualized Sharpe ratio
on a buy-and-hold S&P 500 equity strategy of 0.3-0.4.
At the end of January 2001, the cumulative excess return
since 1986 on the forward-rate bias strategy stood at 233%.
The distribution of monthly returns indicates that the re-
turns are skewed toward positive levels, suggesting posi-
tive investment returns were available from following this
strategy over the 1986-2001 period. High values for rolling
Sharpe ratios lend additional support to the long-run at-
tractiveness of this strategy.
Risk/Reward Structure
of Forward Discount Bias Strategy
(Going Long All High-Yield Currencies and Short All Low-Yield
Currencies Relative to the U.S. Dollar)
(January 1986-April 2002)
Annual Excess Return 5.6%
Standard Deviation 7.8%
Sharpe Ratio 0.73
Note: Datastream is the source of the underlying data.
Cumulative Excess Returns of US$-Based
Forward-Rate Bias Strategy
(1986-2002)
50
100
150
200
250
1986 1988 1990 1992 1994 1996 1998 2000 2002
(Index, 1986=100)
Distribution of Monthly Excess Returns
of US$-Based Forward Rate Bias Strategy
(1986-2002)
0
10
20
30
40
50
-8.2 -7.1 -6.0 -5.0 -3.9 -2.8 -1.8 -0.7 0.3 1.4 2.5 3.5 4.6 5.0
Frequency
Monthly Excess Return(%)
Sharpe Ratios of US$-Based
Forward Rate Bias Strategy
(1986-2002)
-4
-2
0
2
4
6
1986 1988 1990 1992 1994 1996 1998 2000 2002
Five-Year Rolling Sharpe Ratio----
One-Year Rolling Sharpe Ratio____
Forward-Rate Bias Strategy Long-Run Track Record for US$-Based Investors
Annual Excess Percentage Returns
Generated by the DB Forward-Rate Bias Strategy
From the Perspective of U.S. Dollar, Euro, and Yen-Based Investors
(1986-2001)
------ Investors’ Base Currency ------
Year US$ Euro Yen
(%) (%) (%)
1986 -8.4 -7.2 -7.3
1987 -0.6 -0.1 0.1
1988 8.4 8.9 8.9
1989 6.6 7.1 7.2
1990 9.0 8.9 9.3
1991 8.5 7.6 8.2
1992 5.4 4.1 5.1
1993 -4.7 -5.4 -4.5
1994 5.4 5.3 5.5
1995 8.5 9.5 10.4
1996 24.2 24.6 24.7
1997 5.2 5.7 5.8
1998 -10.7 -10.2 -9.2
1999 4.4 4.8 4.7
2000 3.2 3.3 3.5
2001 9.5 9.6 9.8
Note: Datastream is the source of the underlying data.
Source: Datastream
Source: Datastream
Source: Datastream
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 75
Empirical evidence indicates that in the very long run, the
cumulative returns on dollar-denominated and foreign-cur-
rency-denominated short-term fixed-income instruments
are broadly the same. That is, high-yield markets have seen
their currencies weaken over time, while low-yield mar-
kets have seen their currencies strengthen over time.
Indeed, consider the economic consequences if the cu-
mulative returns on a persistently high-yielding market
exceeded the cumulative returns on a persistently low-
yielding market. One market could consistently outperform
another market only if (1) its real yields were consistently
higher, which would dampen that country's long-run growth
prospects, or (2) its currency became progressively more
overvalued on a PPP basis, which would dampen the
country's long-run trade competitiveness. Eventually, a de-
terioration in growth and/or competitiveness would give
rise to corrective moves on the interest-rate and/or ex-
change-rate fronts that would erase the cumulative excess
returns. As shown below, the cumulative returns of U.S.
and foreign money-market instruments in U.S. dollar terms
were virtually the same for the 1990-1999 period.
How, then, do forward-rate bias strategies generate such
attractive risk-adjusted excess returns? The answer is that
in the industrialized world, individual markets often trade
places with other markets in terms of which market has
the highest or lowest yield. When currency
A
's yield rises
above currency
B
's yield, the widening of the
A-B
yield
spread will likely contribute to a rise in currency
A
's value.
When currency
A
's yield begins to fall relative to currency
B
's yield, downward pressure on currency
A
is likely to
materialize. The forward-rate bias strategy would gener-
ate excess returns if an investor went long currency
A
when
its yields rose above currency
B
's, and would also gener-
ate excess returns if that investor went short currency
A
when its yields fell below currency
B
's, as long as the ex-
change rate moved in sympathy with the
A-B
yield spread.
Thus, the forward-rate strategy could generate significant
excess returns, even though the cumulative returns on
currencies
A
and
B
might be the same in the long run.
World Money-Market Performance
(in US$ terms)
80
100
120
140
160
180
200
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000
U.S.----
(Index) Non-US____
Source: Salomon Bothers Money Market Indices
Forward-Rate Bias—The Economic Rationale
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
76 Deutsche Bank Foreign Exchange Research
The monetary, fiscal, external, and financial-sector shocks
that hit Japan in 1995-98 were all largely yen-negative. As
the yen steadily weakened, investors became increasingly
confident that the yen's path was essentially a one-way
street, which encouraged fund managers to take on ag-
gressive long-dollar/short-yen positions. As the yen weak-
ened on a trend basis for 3½ straight years, consistently
remaining above its 60-week moving average, and with
the U.S./Japan short-term yield spread averaging around
500 basis points, the "Yen Carry Trade" (borrowing in yen
to invest in dollar-denominated securities) became enor-
mously popular.
The yen carry trade generated large and persistent excess
returns over the April 1995-August 1998 period. While it is
difficult to quantify precisely how large the outstanding
positions were at their peak, BIS data on foreign-exchange
market turnover suggest that the yen carry trades were
probably quite sizable, given that the yen's share of total
foreign-exchange turnover rose substantially between 1995
and 1998. As the BIS notes, "Between 1995 and 1998, the
value of transactions involving the Japanese currency (at
constant exchange rates) had grown at twice the rate of
those involving the U.S. dollar or the Deutschemark, and
60% faster than global market turnover."
Excessive optimism or pessimism can often carry an ex-
change rate to extreme levels, and such was the case for
the yen. The popularity of the yen carry trade had pushed
the dollar into substantial overbought territory versus the
yen by the summer of 1998.
Undervalued and oversold, the yen was, in retrospect, ripe
for a corrective move upward. Although Japan's fundamen-
tal backdrop had not improved much, the existence of large
highly leveraged positions in the yen carry trade by the
speculative community raised the risk that a sudden un-
winding of those positions could drive the yen's value up
sharply. When the Russian debt crisis hit in the summer of
1998, investors around the world felt compelled to unwind
all highly leveraged fixed-income and currency positions,
including the yen carry trade.
U.S./Japanese Short-Term Interest-Rate Spread
(Three-Month Euro-Deposit Rates)
(1995-2002)
100
200
300
400
500
600
700
1995 1996 1997 1998 1999 2000 2001 2002
(Basis Points)
During the ensuing melee, the yen carry trade became a
victim of the global deleveraging process. The dollar fell by
9% versus the yen between August 31 and September 7,
1998, and then fell an additional 12% October 6-8 as highly
leveraged long-dollar/short-yen positions were aggressively
unwound. From a peak of ¥/US$ 147 on August 11, 1998,
the dollar had fallen by 34 big figures in just two months to
a low of ¥/US$ 113 on October 16, 1998.
Huge losses were incurred during the dollar sell-off of 1998,
which highlights the risks of taking on leveraged positions
in carry trades. However, from a long-run perspective, the
cumulative excess returns on the yen-carry strategy have
been fairly attractive, despite those losses. Indeed, there
have been only two periods (the shaded areas in the chart
below) when the long-dollar/short-yen carry trade has suf-
fered brief, albeit large, short-run losses since April 1995.
Note that the cumulative excess returns from following
this strategy have now surpassed those recorded just be-
fore the unwinding of the yen carry trade in 1998.
Forward-Rate Bias Strategy Opportunities and Pitfalls—
The Case of the Yen Carry Trade
Source: Datastream
Source: Datastream
Source: Datastream
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 77
Conventional wisdom holds that countries that run persis-
tent current-account surpluses will see their currencies
appreciate over time, while countries that run persistent
current-account deficits will see their currencies depreci-
ate over time.
Current-account imbalances can affect exchange rates
through a variety of channels. First, the existence of major
external imbalances can influence the flow supply of and
demand for individual currencies and thereby directly in-
fluence the paths that exchange rates take.
Second, major current-account imbalances can shift the
residence of financial wealth among deficit and surplus
nations, decreasing it in the former and raising it in the
latter. Such shifts in the residence of financial wealth could
then lead to a shift in global asset preferences, which, in
turn, could influence the paths that exchange rates take.
Third, there is a general notion that there exists some limit
on the ability of countries to run sizeable and persistent
current-account deficits, because such deficits could lead
to an unending rise in debt owed by deficit countries to
foreign investors. If foreign investors view the rise in ex-
ternal debt as unsustainable, they would reason that at
some point, a major depreciation of the deficit country's
currency might be required to ensure that the current-ac-
count deficit narrowed and that the external debt stabi-
lized at a level deemed sustainable.
In such a manner, the existence of a large current-account
imbalance will tend to alter the market's notion of what
level of the exchange rate constitutes a currency's real long-
run equilibrium value, with ever-larger deficits and external
debt levels giving rise to steady, downward revisions in
market expectations of the real long-run equilibrium ex-
change rate.
Current Account Imbalances and Exchange Rates—
The Channels of Influence
Flow
Supply and Demand
for Foreign Exchange
Exchange
Rate
Transfer of Wealth
from Deficit
to Surplus Countries
External
Debt
Sustainability
Current
Account
Imbalance
Current-Account Imbalances and the Determination of Exchange Rates
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
78 Deutsche Bank Foreign Exchange Research
Most market participants view U.S. current-account trends
as important for the dollar because such trends directly
influence the flow supply of and demand for dollars in the
foreign-exchange market. Using the traditional balance-of-
payments flow model of exchange-rate determination, we
can describe how current-account flows determine the
supply of and demand for dollars.
Let's assume that the flow demand for dollars, D$, is gen-
erated by foreign demand for U.S. goods and services,
while the flow supply of dollars, S$, is generated by U.S.
demand for foreign goods and services. If current-account
flows determine the supply of and demand for dollars, then
it should be the case that when the U.S. current-account
balance is in equilibrium, the supply of and demand for
dollars will be in equilibrium as well. In terms of the dia-
gram below, the dollar's equilibrium value, q0, would be
determined by the intersection of the supply, S$, and de-
mand, D$, curves for dollars.
Balance-of-Trade Flow Model
of Exchange-Rate Determination
Dollar’s Value
Quantity of
Dollars
D
$
S
$
q
0
AB
q
1
Q
A
Q
B
C
If, however, the U.S. were running a current-account defi-
cit, perhaps because the value of the dollar was too high
at q1 relative to its equilibrium value, q0, then the supply of
dollars, S$, on the foreign-exchange market would exceed
the demand for dollars, D$. The excess supply of dollars,
which is the foreign-exchange market counterpart of the
U.S. current-account deficit, is represented as the gap be-
tween points
A
and
B
. The dollar's value at q1 would clearly
not be sustainable since the excess supply of dollars,
AB
,
on the foreign-exchange market would exert continual
downward pressure on the dollar's value. Once the dollar's
value falls toward its long-run equilibrium value, q0, the
supply of dollars would be brought back into line with the
demand for dollars, and the U.S. current account would be
brought back into balance at point
C
.
Balance of Payments Flow Model of Exchange-Rate Determination
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 79
The traditional balance-of-payments flow model posits that
exchange rates will adjust until all current-account imbal-
ances are corrected and current-account equilibrium is re-
stored. How much exchange rates must adjust to restore
current-account equilibrium depends on the elasticities of
the supply and demand curves for foreign exchange.
In the accompanying diagram, we plot two sets of supply
and demand curves for dollars—one set that is highly elas-
tic (Se$ and De$, respectively) and another set that is highly
inelastic (Si$ and Di$, respectively). Initially, it is assumed
that the U.S. is running a current-account deficit equal to
AB
at the prevailing exchange rate, q1. The exchange-rate
change necessary to correct the U.S. current-account im-
balance will differ depending on whether the prevailing
supply and demand curves for dollars are highly elastic or
inelastic.
Elasticities of Supply and Demand
for Dollars and
Exchange-Rate Determination
Exchange Rate
Quantity of
Dollars
D
e
$
AB
Q
A
Q
E
Q
B
q
1
S
e
$
S
i
$
D
i
$
C
q
0
q
2
If the supply and demand curves for dollars were highly
elastic, only a modest decline in the dollar's value to q0
would be required to restore equilibrium to the U.S. cur-
rent account. If, instead, the supply and demand curves
for dollars were highly inelastic, a large decline in the dol-
lar to q2 would be required to restore equilibrium to the
U.S. current account.
U.S. export and import price elasticities are estimated to
be quite small, perhaps averaging only around -1.0 for ex-
ports and -0.3 for imports. Based on these rather low elas-
ticities, it would likely require a rather large depreciation of
the dollar to restore a sizeable U.S. current-account deficit
to a more sustainable level. Thus, the inelastic supply and
demand curves shown below may be a more realistic pic-
ture of the potentially large dollar adjustment that might
be required to restore equilibrium to the U.S. current-ac-
count balance.
Trade Elasticities and the Exchange-Rate/Trade-Balance
Adjustment Process
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
80 Deutsche Bank Foreign Exchange Research
The dramatic appreciation of the dollar over the 1995-2001
period despite a record deterioration of the U.S. current
account is clear evidence that a current-account deficit need
not be bearish for a currency. Indeed, the same thing hap-
pened in the first half of the 1980s when the dollar soared,
yet the current-account deficit steadily widened.
Current-account imbalances are typically driven by struc-
tural changes in international competitiveness, changes in
the terms of trade, and long-term shifts in national sav-
ings-investment (S-I) balances. Consider the impact on
exchange rates and the current account if a country under-
went a major investment boom or undertook a long-term
policy of fiscal expansion. In both cases, the S-I schedule
would shift upward to the left. The increase in investment
relative to national savings would result in a wider current-
account deficit and, at the same time, put upward pres-
sure on the real exchange rate. That is pretty much what
happened to the U.S. savings-investment balance, the cur-
rent-account balance, and the dollar in both the 1980-85
and 1995-2001 episodes. Hence, when the source of the
deterioration in the current-account balance is a rise in in-
vestment relative to national savings, there is a high likeli-
hood that the currency will strengthen over time, not
weaken.
Current-Account/Exchange-Rate Linkages
and the Impact of Economic and Financial Disturbances
Economic Current Exchange
Conditions Account Rate Linkage
Expansionary Monetary Policy Deterioration Depreciation Positive
Restrictive Monetary Policy Improvement Appreciation Positive
Expansionary Fiscal Policy Deterioration Appreciation Negative
Restrictive Fiscal Policy Improvement Depreciation Negative
Exogenous Increase Improvement Appreciation Positive
in Domestic Supply
Exogenous Increase Improvement Appreciation Positive
in Foreign Demand
U.S. Savings-Investment Balance
and the Dollar
Savings minus Investment,
Current-Account Balance
U.S. Dollar
Real Exchange Rate
“New Economy”
Increase in
Investment
Spending
S-I
2
CAB
1
“New Economy”
Upward Revision in the
Dollar’s Real Long-Run
Equilibrium Value
q
2
q
1
(0)(-) (+)
S-I
1
If the deterioration in a country's current-account balance
were instead due to a long-run deterioration in that
country's competitiveness, the CAB schedule would shift
downward to the left relative to an unchanged S-I sched-
ule. In such a case, the deterioration of the current ac-
count would coincide with a long-term slide in the domes-
tic currency’s real long-run equilibrium value.
The message from all this is that certain economic distur-
bances may cause exchange rates and the current-account
balance to move in the same direction while other distur-
bances may cause exchange rates and the current account
to diverge. Consider the current-account/exchange-rate link-
age that arises in response to fiscal and monetary policy
changes. A country that pursues a highly expansionary
monetary policy will likely see its current account deterio-
rate and its currency weaken at the same time. The ques-
tion is whether the weakening of the currency is due to
the easier monetary policy or the current-account deterio-
ration or possibly both. In the case of a highly expansion-
ary fiscal policy, the trend in the current account and the
trend in the exchange rate may diverge as the current ac-
count deteriorates while the currency appreciates.
Are Current-Account Deficits Necessarily Bearish for Exchange Rates?
U.S. Current-Account Deficit and the Dollar
Savings minus Investment,
Current-Account Balance
U.S. Dollar
Real Exchange Rate
S-I
CAB
2
CAB
1
Downward Revision in the
Dollar’s Real Long-Run
Equilibrium Value
Long-Run Deterioration in
U.S. Trade Competitiveness
q
1
q
2
(0)(-) (+)
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 81
Japan's tendency to run large and persistent current-ac-
count surpluses over time largely explains why the yen's
trade-weighted value has risen on a secular basis. From a
simple flow supply-and-demand perspective, one would
expect larger and larger surpluses to give rise to succes-
sive increases in the demand for yen over time. Indeed,
the rising trend in Japan's current-account surplus has
moved in sympathy with the rising trend in the yen's value
in the past 30 years.
Related to this is the fact that Japan's cumulative current-
account surpluses have allowed Japanese investors to
acquire a rising claim on U.S. assets. To induce Japanese
investors to acquire and hold onto those rising dollar claims,
either U.S. interest rates needed to rise relative to those in
Japan, or the dollar needed to fall versus the yen to make
dollar assets appear cheaper to Japanese investors, or
some combination of the two was required. The evidence
indicates the dollar bore the brunt of this adjustment. The
accompanying chart shows that the yen’s long-term trend
moved sympathetically with the ratio of Japan's interna-
tional investment position relative to the U.S. international
investment position, as a percentage of their respective
GDPs.
In addition to the long-term financial pressure on yen ema-
nating from Japan's persistent current-account surpluses,
Professor Ronald I. McKinnon of Stanford University makes
the case that commercial tensions between the U.S. and
Japan, including intermittent threats of an outright trade
war, may have played an important role in driving the yen's
real value higher over time. According to McKinnon, the
U.S.—concerned about Japan's large and rising bilateral
trade surplus and frustrated by the inability of U.S. firms to
penetrate Japan's relatively closed markets—pursued a
policy of coupling protectionist threats with demands, im-
plicit or explicit, for yen appreciation. With the yen the fo-
cus of both U.S. trade and exchange-rate policies, the Japa-
nese authorities had to either tolerate an ever-rising yen or
accept the wrath of American protectionists.
Repeated attempts to talk the dollar down versus the yen
created, according to McKinnon, a syndrome of an ever-
rising yen that had a pervasive influence on investor ex-
pectations in Japan. Concerned about possible large losses
on their dollar-based investments, brought about by re-
peated attempts by U.S. officials to talk the dollar down,
Japanese investors became increasingly reluctant to hold
too large a proportion of their total portfolio in foreign as-
sets. This meant that Japanese funds would effectively be
bottled up inside Japan, with no safe-passageway for
Japan's current-account surpluses to exit through. This
contributed to an ever-rising basic balance of payments
surplus in Japan (the current account plus the long-term
capital account). As shown in the accompanying chart, the
tendency of Japan's basic balance of payments to regis-
ter a persistent surplus over time played a key role in driv-
ing the yen higher over time.
The Japanese Yen and the U.S./Japan
Cumulative Current-Account/GDP Differential
(50)
(40)
(30)
(20)
(10)
0
10
20
50
100
150
200
250
300
77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96
Cum. Curr.-Acct. as % of GDP Diff.---- (%)
¥
/US$____
The Yen and Japan’s Current-Account Balance
-10
-5
0
5
10
15
20
(350)
(300)
(250)
(200)
(150)
(100)
(50)
76 78 80 82 84 86 88 90 92 94 96 98 00 02
Current Acc’t(¥ tr.)----
Y
en/US$(reverse scale)____
The Yen and Japan’s Basic Balance
-10
-5
0
5
10
60
80
100
120
140
160
180
86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02
Basic Balance(12-mo. moving sum)(¥ tr.)----
Y
en Index____
Japan’s Current-Account Balance and the Yen
Source: Datastream
Source: Datastream
Source: Datastream
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
82 Deutsche Bank Foreign Exchange Research
There are essentially three ways that the U.S. can narrow
its large current-account deficit. First, the U.S. could pur-
sue policies that encourage domestic savings and/or dis-
courage domestic investment, perhaps via a tighter fiscal
and/or tighter monetary policy. Such policies would con-
tribute to a leftward shift of the savings-investment (S-I)
schedule in the diagram on the right. As illustrated, U.S.
domestic demand would need to be pared back, perhaps
significantly, to restore sustainable balance to the U.S. ex-
ternal account.
A second way to eliminate the unsustainable portion of
the U.S. current-account deficit would be to engineer a
major depreciation of the dollar to improve U.S. competi-
tiveness, and thereby boost net exports. A major dollar
depreciation would shift the export-import (Ex-Im) sched-
ule to the right to Ex-Im1 and would to narrow the deficit
from point
A’
to a more sustainable level such as point
B’
.
An even larger dollar depreciation would be needed to shift
the export-import curve to Ex-Im2 to narrow the deficit to
point
C’
.
The problem with this strategy is that it might create other
problems for U.S. policymakers, and in all likelihood, might
fail completely. That is because if the U.S. economy were
already operating at its full employment level of output Y'F,
then a dollar depreciation that shifts the export-import curve
to the right would drive the level of U.S. output past its full
employment potential. Since this would be highly inflation-
ary, the Federal Reserve would need to tighten monetary
policy, perhaps significantly. But if the Federal Reserve
were forced to push U.S. interest rates up sharply, that
would attract capital inflows to the U.S., which would then
strengthen the dollar, thereby offsetting, perhaps com-
pletely, the initial depreciation of the dollar.
A third way to eliminate the unsustainable portion of the
U.S. current-account deficit would be to combine belt-tight-
ening and dollar-depreciation initiatives. A modest belt-tight-
ening move on the fiscal and monetary policy front could
shift the S-I curve to the left to S-I3, while a modest dollar
depreciation could work to shift the export-import curve to
the right to Ex-Im3. The combination of these two policy
actions would leave the equilibrium level of output un-
changed while narrowing the U.S. current-account deficit
to a sustainable level.
Correcting the U.S. Current-Account Deficit—
Belt Tightening vs. Dollar Depreciation
Correcting the U.S. Current-Account Imbalance
via a Weakening of the Dollar
Output
Saving-Investment,
Exports-Imports
C
B’
C’
S-I
0
Ex-Im
2
Ex-Im
1
A’
B
A
Y’
F
Y’’ Y
Ex-Im
0
0
Correcting the U.S. Current-Account Imbalance
via Tighter Fiscal/Monetary Policies
plus a Weakening of the Dollar
Output
Saving-Investment,
Exports-Imports
C
C’
S-I
3
S-I
0
Ex-Im
3
A’
A
Y’
F
Ex-Im
0
0
Correcting the U.S. Current-Account Imbalance
via Tighter Fiscal and Monetary Policies
Output
Saving-Investment,
Exports-Imports
C
B’
C’
S-I
0
S-I
2
S-I
1
Ex-Im
0
A’
B
A
Y’
F
Y’Y’
0
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 83
Many pundits have argued that the record deterioration of
the U.S. current-account deficit will inevitably lead to a hard
landing for the dollar. So far it has not.
One possible explanation for the dollar’s resiliency is that
the current-account deterioration might be an
equilibrium
and not a
disequilibrium
phenomenon. If one views U.S.
economic developments from a "New Economy" perspec-
tive, the revolutionary changes in information technology
may have raised both the speed limit on sustainable U.S.
GDP growth and the sustainable U.S. current-account defi-
cit.
With the return on capital on new technology investments
highly attractive, foreign investors have been more than
willing to finance a larger-than-normal gap between U.S.
savings and investment. As a result, the inflow of foreign
capital into the U.S. has exceeded the typical flow of capi-
tal that found its way to the U.S. in the past. Since, by
definition, a country's current-account deficit equals its
capital-account surplus, which also equals its savings-in-
vestment gap, it would appear that the information tech-
nology revolution in the U.S. simultaneously raised both
the speed limit on sustainable U.S. growth and the size of
the sustainable current-account deficit that the U.S. can
safely run without triggering a dollar crisis. In fact, the in-
formation technology revolution, which is presently domi-
nated by U.S. firms, may have lifted the dollar's real long-
run equilibrium level as well.
While it might be true that the size of the sustainable cur-
rent-account deficit has risen, there nevertheless must exist
some threshold level for the U.S. current-account deficit
that would trigger a decline in the dollar’s value. The key
questions for investors then are: (1) what might the new
higher threshold level be, and (2) how soon will it be be-
fore the U.S. current-account deficit crosses that thresh-
old and begins to take its toll on the dollar?
U.S. Dollar Index
75
80
85
90
95
100
105
110
115
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
US$ Major Currency Index
U.S. Current-Account Balance
-500
-400
-300
-200
-100
0
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
(US$ bn.) Annualized
U.S. Current-Account Deficit and the Dollar
Source: Datastream
Source: Datastream
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
84 Deutsche Bank Foreign Exchange Research
Based on an analysis of 17 episodes in industrial countries
over the 1980-95 period, Catherine L. Mann (1999) found
that current-account deficits reaching 4.2% of GDP tended
to set forces in motion—including corrective currency ad-
justments—for a reversal of the current-account shortfall.
The U.S. current-account deficit as a percentage of GDP is
presently at 4.1%. If the U.S. economy recovers strongly
over the course of 2002, the U.S. current-account deficit
could widen to as much as 5% of GDP, well above the
4.2% average threshold.
Another study (Caroline L. Freund, 2000) examined a larger
sample of industrial country current-account adjustment
episodes and found that the median current-account defi-
cit/GDP ratio that triggered corrective adjustments in the
current account was around 4.9%. Those results are re-
ported in the table below.
As shown by the median current-account adjustment pro-
cess, an industrial-country deficit of 4.9% of GDP has, in
the past, led to a 19% real depreciation of the deficit
country's currency over a three-year period, with the de-
cline beginning one year prior to the peak in the current-
account deficit and ending about three years after the peak.
In conjunction with a slowdown in domestic demand, the
decline of the currency has typically helped bring about an
83% decline in the current-account imbalance. If the U.S.
conforms to the median episode, and the U.S. current-ac-
count deficit/GDP ratio rises to 5% by year-end 2002, then
the corrective downward adjustment in the dollar's value
(which typically occurs one year prior to the peak in the
deficit) should have started in early 2002.
Estimating the Threshold Level for the U.S. Current-Account Deficit
That Would Trigger a Decline in the Dollar’s Value
Current-Account Adjustment and Real Exchange-Rate Depreciation
(in 25 Episodes of Large Current-Account Adjustments, 1980-1995)
Relative to Peak Deficit Year (t0),
Current Account/GDP Ratio Three-Year Real the Year in which
at Peak 3 Years after Peak Percentage Exchange-Rate Exchange-Rate Depreciation
Date (t0) (t
3) Change Depreciation Began Ended Duration
Australia (1989) -6.2% -3.7% 40% 21% 1 5 4
Austria (1980) -4.9% 0.4% 108% 19% -1 2 3
Belgium (1981) -4.2% -0.1% 98% 51% -3 4 7
Canada (1981) -4.2% -0.4% 91% ----
Canada (1993) -3.9% 0.6% 114% 19% -1 3 4
Denmark (1986) -5.3% -1.0% 81% ----
Finland (1991) -5.5% 1.3% 124% 30% -1 2 3
France (1982) -2.2% 0.0% 100% 15% -1 3 4
Greece (1985) -8.1% -1.5% 82% 21% -2 2 4
Hong Kong (1980) -5.0% 0.6% 111% ----
Ireland (1981) -13.6% -5.6% 59% 41% -2 1 3
Israel (1982) -8.7% 4.8% 155% 14% 2 6 4
Italy (1981) -2.6% -0.8% 70% 12% -1 1 2
Italy (1992) -2.5% 2.3% 194% 28% 0 4 4
Korea (1980) -8.5% -1.8% 78% ----
New Zealand (1984) -13.6% -8.0% 41% 11% -1 1 2
Norway (1986) -6.0% 0.2% 104% 5% -3 1 4
Portugal (1981) -16.8% -2.8% 83% 8% 1 3 2
Singapore (1980) -13.3% -3.5% 74% ----
Spain (1981) -2.9% 1.1% 139% 36% -1 3 4
Spain (1981) -3.8% -1.4% 62% 17% 1 4 3
Sweden (1980) -3.5% -0.8% 77% 20% 1 4 3
Sweden (1992) -3.6% 2.1% 160% 21% 1 4 3
United Kingdom (1989) -4.4% -1.8% 60% 15% 3 5 2
United States (1987) -3.7% -1.7% 56% 34% -1 2 3
Median -4.9% -0.8% 83% 19% -1 3 3
Source: Caroline L. Freund, “Current Account Adjustment in Industrial Countries", Federal Reserve Board, International Finance Discussion Paper, No. 692, December 2000.
DB Guide to Exchange-Rate Determination
May 2002 Deutsche Bank @
Deutsche Bank Foreign Exchange Research 85
After the experiences of the early 1980s and the late 1990s,
when the dollar soared despite the significant widening of
the U.S. current-account deficit, one could make the case
that the U.S. current-account deficit might be overrated as
a key determinant of the dollar. In fact, one could question
whether the measurement problems associated with cur-
rent-account data makes them a reliable measure of U.S.
trade competitiveness.
The reported current-account data omit the contribution
that U.S. foreign affiliates make to the aggregate sales by
U.S. firms in international markets. When allowance is
made for the role played by U.S. foreign affiliates, one finds
that U.S. firms' penetration of foreign markets actually ex-
ceeds the penetration of foreign firms in the U.S. market.
The U.S. Commerce Department reports that worldwide
sales by U.S. companies to foreign residents (which in-
clude U.S. exports plus total sales by U.S. companies' for-
eign affiliates to non-residents) amounted to $3.17 trillion
in 1998 (latest data available), while total sales by foreign
companies to U.S. residents (which include U.S. imports
plus total sales by foreign firms' U.S. affiliates to U.S. resi-
dents) amounted to $2.81 trillion in the same period. That
means that worldwide sales by U.S. companies to foreign
residents exceeded the sales of foreign firms to U.S. resi-
dents by $363 billion. Breaking the data down, the Com-
merce Department notes that of the $3.17 trillion in U.S.
firms' worldwide sales, $930 billion represented U.S. ex-
ports of goods and services, while $2.24 trillion repre-
sented the sales by U.S. firms' foreign affiliates. Of the
$2.81 trillion in foreign companies' sales to U.S. residents,
$1.1 trillion represented U.S. imports of foreign goods and
services, while $1.7 trillion represented the sales by for-
eign firms' U.S. affiliates.
What is clear from the data is that although U.S. imports
($1.1 trillion) exceeded U.S. exports ($930 billion), leaving
a wide U.S. trade deficit of $167 billion, the sales of U.S.
firms' foreign affiliates ($2.24 trillion) far exceeded the sales
of foreign firms' U.S. affiliates ($1.71 trillion), leaving a U.S.
surplus on foreign affiliate operations of $530 billion.
Overall, despite the fact that the U.S. runs a large trade
deficit, U.S. firms have had greater success in terms of
penetrating foreign markets than foreign firms have had in
penetrating the U.S. market. The data show that, at the
margin, U.S. firms tend to accommodate foreign demand
through their overseas affiliates (71% of worldwide sales)
rather than through outright exports from the U.S. (only
29% of worldwide sales). In contrast, foreign firms, at the
margin, tend to rely more on exports to the U.S. (39% of
their U.S. sales) and a bit less on their U.S.-based affiliates
(61% of their U.S. sales) to accommodate demand in the
U.S. market.
What does this all imply for the U.S. current-account
deficit's impact on the dollar? Does the dollar need to de-
cline in order to balance U.S. exports and imports? Does
the dollar have to be penalized because U.S. firms tend to
accommodate foreign demand more through their over-
seas affiliates than via direct exports, particularly given the
fact that U.S. firms do a better job in the aggregate than
their foreign counterparts in terms of penetrating the
other's market? Or should the dollar be rewarded to re-
flect the global reach and strength that U.S. firms exhibit
in the market?
Worldwide Sales by U.S. Companies versus
Foreign Companies’ Sales to the U.S.
(Latest Data, 1998)
(US$ Billions)
Worldwide Sales Sales by Foreign
by U.S. Firms Firms to the U.S. Difference
Total $3,173 $2,810 +$363
Sales by
Foreign Affiliates $2,240 (71%) $1,710 (61%) +$530
Exports $933 (29%) $1,100 (39%) -$167
Source: U.S. Commerce Department
Is the U.S. Current-Account Balance a Reliable Measure of
U.S. Trade Competitiveness?
DB Guide to Exchange-Rate Determination May 2002
Deutsche Bank @
86 Deutsche Bank Foreign Exchange Research
Greater financial integration of the world's capital markets
and the increased freedom of capital to flow across na-
tional borders have increased the importance of financial
flows in the determination of exchange rates. The rise in
the dollar's value in recent years was largely driven by
strong increases in the demand for U.S. assets by over-
seas investors. For example, Euroland investors had been
structurally underweight equities for a long time, but be-
ginning in the late 1990s they started to shift their portfo-
lio allocations away from bonds and toward equities. Within
their rising equity holdings, European investors also made
the decision to increase the percentage allocated to U.S.
equities and reduce the percentage allocated to Euroland
equities. Overall, there was a marked rise in European in-
vestors' demand for U.S. equities in 1999-2001 that helped
lift the dollar versus the euro.
Foreign demand for Japanese equities has also played a
role in influencing the medium-term direction of the yen.
The yen's weakness over the 1995-98 period coincided
with declining foreign demand for Japanese equities, and
the yen's subsequent rebound in late 1998 and in 1999
coincided with a rise in