PRINCETON STUDIES IN INTERNATIONAL ECONOMICS

No. 89, May 2001

THE MICROSTRUCTURE
OF THE FOREIGN-EXCHANGE MARKET:
A SELECTIVE SURVEY OF THE LITERATURE

LUCIO SARNO
AND
MARK P. TAYLOR

INTERNATIONAL ECONOMICS SECTION

DEPARTMENT OF ECONOMICS
PRINCETON UNIVERSITY
PRINCETON, NEW JERSEY

PRINCETON STUDIES
IN INTERNATIONAL ECONOMICS

PRINCETON STUDIES IN INTERNATIONAL ECONOMICS are

published by the International Economics Section of the
Department of Economics of Princeton University. Although the Section sponsors the Studies, the authors are
free to develop their topics as they wish. The Section
welcomes the submission of manuscripts for publication
in this and its other series. Please see the Notice to
Contributors at the back of this Study.
The authors of this Study are Lucio Sarno and Mark P.
Taylor. Dr. Sarno is Reader (Associate Professor) in
Financial Economics at the Warwick Business School,
Warwick University, and Research Affiliate at the Centre
for Policy Research in London. He has also served at
Brunei University, Columbia University, the University of
Liverpool, and the University of Oxford and has been
involved in research and consulting projects for the
European Commission, the Federal Reserve Bank of St.
Louis, and the World Bank. He has written numerous
articles about the foreign-exchange market, capital
movements, and related subjects.
Dr. Taylor is Professor of Finance at the Warwick
Business School, Warwick University, and Research Fellow
at the Centre for Economic Policy Research. He served for
five years as a Senior Economist at the International
Monetary Fund and has also been an Economist at the
Bank of England and a Consultant for the World Bank. He
is the author of several books and more than one hundred
articles.
GENE M. GROSSMAN, Director
International Economics Section

PRINCETON STUDIES IN INTERNATIONAL ECONOMICS

No. 89, May 2001

THE MICROSTRUCTURE
OF THE FOREIGN-EXCHANGE MARKET:
A SELECTIVE SURVEY OF THE LITERATURE

LUCIO SARNO
AND
MARK P. TAYLOR

INTERNATIONAL ECONOMICS SECTION

DEPARTMENT OF ECONOMICS
PRINCETON UNIVERSITY
PRINCETON, NEW JERSEY

INTERNATIONAL ECONOMICS SECTION

EDITORIAL STAFF
Gene M. Grossman Director
Pierre-Olivier Gourinchas Assistant Director
Margaret B. Riccardi, Editor
Sharon B. Ernst, Editorial Aide
Lalitha H. Chandra, Subscriptions and Orders
Library of Congress Cataloging-in-Publication Data
Sarno, Lucio
The microstructure of the foreign-exchange market: a selective survey of the literature
/ Lucio Sarno and Mark P. Taylor.
p. cm. (Princeton studies in international economics ; no. 89)
Includes bibliographical references.
ISBN 0-88165-261-X
1. Foreign-exchange market. I. Taylor, Mark P, 1958. II. Title. III. Series.
HG3851.S27 2001
332.45dc21

2001024516
CIP

Copyright 2001 by International Economics Section, Department of Economics,

Princeton University.
All rights reserved. Except for brief quotations embodied in critical articles and reviews,
no part of this publication may be reproduced in any form or by any means, including
photocopy, without written permission from the publisher.
Printed in the United States of America by Princeton University Printing Services at
Princeton, New Jersey
International Standard Serial Number: 0081-8070
International Standard Book Number: 0-88165-261-X
Library of Congress Catalog Card Number: 2001024516

International Economics Section

Department of Economics, Fisher Hall
Princeton University
Princeton, New Jersey 08544-1021

Tel: 609-258-4048
Fax: 609-258-1374
E-mail: [email protected]
Url: www.princeton.edu/~ies

CONTENTS
1

INTRODUCTION

THE FOREIGN-EXCHANGE MARKET

3
3
9

Institutional Features of the Foreign-Exchange Market

Foreign-Exchange-Market Activity
3

EXCHANGE-RATE EXPECTATIONS

Survey-Date Studies
Order-Flow Studies
4

CHARTIST ANALYSIS

Chartist Analysis in the Foreign-Exchange Market

The Impact of Chartist Analysis
Modeling Expectations of Chartists, Fundamentalists,
and Portfolio Managers
Summing Up
5

TIME-VARYING VOLATILITY, MARKET LOCATION, AND TRADING

VOLUME

Time-Varying Volatility in the Foreign-Exchange Market

Volatility and Market Location
Public Information or Private Information?
6

BID-ASK SPREAD DETERMINATION

The Main Determinants of the Spread

Modeling the Spread: Theory and Evidence
7

MARKET MAKERS BEHAVIOR

11
11
12
14
14
16
17
19

21
21
24
27
33
33
34

Private Beliefs and Information Externalities

"Event Uncertainty" versus "Hot Potato"

40
40
41

CONCLUSION

43

REFERENCES

44

1 INTRODUCTION
Over the last quarter century, exchange rates among the currencies of the
leading industrial countries have shown substantial and often persistent
movements that are largely unexplained by movements in macroeconomic
fundamentals (Frankel and Rose, 1995; Taylor, 1995; Flood and Taylor,
1996). The literature on the microstructure of the foreign-exchange market
in some measure reects researchers attempts to understand the mechanisms generating these deviations from fundamentals (Taylor, 1995; Flood
and Taylor, 1996; Lyons, 2001).1 More broadly, however, this literature is
concerned with other micro aspects of the foreign-exchange market, such
as the transmission of information among market participants, the behavior
of market agents, the importance of order ow, the heterogeneity of agents
expectations, and the implications of such heterogeneity for trading volume
and exchange-rate volatility.
The assumptions and methodology of the microstructure literature often
dier substantially from those of the conventional macroeconomic approach.
Indeed, some authors see the microstructure literature as a radical departure from the traditional modelling strategy of treating foreign exchange
rates as a macroeconomic relative price (Frankel and Rose, 1995. p. 1710).
With respect to assumptions, the foreign-exchange microstructure typically
does not assume that only public information is relevant to exchange rates,
that foreign-exchange-market agents are homogeneous, or that the mechanism used for trading is inconsequential (Lyons, 2001). In fact, these issues
are themselves often the subject of investigation in microstructural analysis. With respect to methodology, instead of starting with a set of macroeconomic relations such as money demand and purchasing-power parity,
which are then used to solve for the exchange rate (Taylor, 1995), the microstructure literature analyzes the behavior and interaction of individual
decisionmaking units in the foreign-exchange market.2 Simply put, the microstructure literature is concerned with the details of the mechanics of
1 One way in which a speculative movement of the exchange rate away from the
level consistent with the macroeconomic fundamentals could begin would be if some
agents were to have destabilizing expectationsif, for example, a 5 percent appreciation
were to lead agents to expect a 10 percent appreciation, which would lead them to
buy the appreciating currency, causing it to appreciate faster in a self-fullling fashion.
Bandwagon eects of this kind have been examined by Frankel and Froot (1986), Allen
and Taylor (1990), and others, using survey data on expectations. The evidence to date
tentatively suggests that expectations may be more destabilizing over relatively short
horizons.
2 The microstructure approach should also be distinguished from the dynamic optimizing approach of the new open-economy macroeconomics (Obstfeld and Rogo,
1995, 1996; Lane, 1999; Sarno, 2001). Although the new open-economy literature explic-

foreign-exchange trading, whereas the macroeconomic approach typically

dismisses the details as unimportant.3
Although it is possible to imagine a convergence of the macroeconomic
and microstructural approaches, they currently remain largely separate.4
Indeed, as one pioneering investigator in the area of foreign-exchange microstructure (Lyons, 2001, p. 9) notes:
Researchers working with microstructure tools are not applying them
to issues on the macro end of the issues spectrum and those who are
working on the macro end have not felt it worthwhile to invest in
apparently inappropriate microstructure tools. For these tools to be
applied to more macro foreign exchange issues, either macroeconomists need introduction to the tools or those with the tools need
to extend their study rightward on the spectrum. Both of these, in
fact, are occurring. As the process continues, the uneasy dichotomy
between macro and micro approaches is destined to fade.

For the convergence process to continue, therefore, the microstructure literature needs to be more widely understood. It is in this spirit that the
present survey is oered.
The remainder of this study is set out as follows. Chapter 2 gives a general overview of the institutional features of the foreign-exchange market.
Chapter 3 discusses studies employing survey data on market participants
exchange-rate expectations and studies using data on foreign-exchange order ow. Chapter 4 reviews the literature on chartist, or technical,
analysis. Chapter 5 summarizes recent work on modeling time-varying
volatility in the foreign-exchange market, on the relation between volatility
and trading volume, and on the nature of information that is relevant to
trading in the foreign-exchange market. Chapter 6 reviews the literature
on bid-ask spread determination in the foreign-exchange market. Chapter
7 discusses the literature on modeling market makers behavior. A nal
chapter summarizes and concludes the study.
itly takes an optimizing approach (usually of representative agents in two economies),
rather than relying on o the peg macroeconomic behavioral relations, it is generally concerned with interactions between whole economies, rather than strictly with the
functioning of the foreign-exchange market.
3 Flood and Taylor (1996, p. 285), for example, write: It is apparent that there
are important inuences, not on the list of standard macro fundamentals, that aect
short-run exchange rate behavior . . . , and it is in this context that new work on the
microstructure of the foreign exchange market seems both warranted and promising.
4 Researchers have, in the past, occasionally employed methods of analysis that eectively synthesize macroeconomic and microeconomic approaches. For example, Taylors
(1987, 1989) studies of covered interest parity may be viewed as early microstructural
analysis, applying high-frequency, high-quality data from the brokered foreign-exchange
market and microstructural tools to resolve an issue (the validity of covered interest
parity) debated up till now only at the macroeconomic level.

2 THE FOREIGN-EXCHANGE MARKET

Institutional Features of the Foreign-Exchange Market
Much of the early microstructure literature focuses on institutional features
of the foreign-exchange market. The rst section of this chapter briey
discusses these features; the second section presents some simple data on
the activity of real-world foreign-exchange markets.
The two main features distinguishing the foreign-exchange market from
other nancial markets are that the foreign-exchange market is a mostly decentralized market and that trading volume in the foreign-exchange market
is very large and conducted mainly among market makers.1 A decentralized
market is a market in which participantsnamely, market makers, brokers,
and customers2 are generally physically separated from each other and in
which transactions are made by telephone, telex, or computer networks.
In a centralized market, trade is carried out at a publicly announced price
and all traders in the market face the same trading opportunities. In a
decentralized system, prices are quoted and transactions executed in private meetings, perhaps conducted through some form of electronic medium
(Schwartz, 1988; Wolinsky, 1990). The New York Stock Exchange (NYSE)
and the London Stock Exchange (LSE), for example, are centralized markets, but the direct foreign-exchange market may be classied as decentralized, and the brokered market may be classied as quasicentralized,
because each foreign-exchange broker accumulates a subset of market makers limit orders.3
1 Another institutional feature of the foreign-exchange market is that a large fraction
of the trading volume in the market is concerned with the trade of forward contracts
(Flood, 1991; BIS, 1998). Forward positions (that is, contracts concerning the future exchange of currencies at pre-agreed rates) may be taken by agents engaged in international
trade in goods and services, in order to hedge against adverse exchange-rate movements
aecting their business, and by dealers and speculators hoping to earn windfall profits from correctly anticipating exchange-rate movements. See the following section and
footnote 12 of this chapter for a discussion of forward contracts in the foreign-exchange
market.
2 Broadly speaking, a market maker is a rm (typically a bank) attached to the foreignexchange market and engaged in the buying and selling of currencies; it thus establishes a
market for these currencies by standing ready to buy and sellthat is, to make a twoway price forparticular currencies against one another. Such rms generally make
signicant prots from the dierence between the bid exchange-rate price at which
they buy a currency and the (higher) ask price at which they sell. The market maker
may be seen as the principal in the buying and selling of currencies, whereas the broker
acts as an agent on behalf of customers wishing to buy or sell.
3 A limit order is an oer by a bank to buy or sell, but not both, a certain amount of
currency against another currency at a specied rate of exchange.

The empirical evidence from the current microstructure literature strongly suggests that the dierent degrees of centralization across markets may be
very important in explaining dierences in market performance. A widely
held view is that, by centralizing trading and price information, the brokered market uses time more eciently, eliminates signicant arbitrage opportunities rapidly, and ensures dealers that orders are executed according
to price priority (Garbade, 1978; Garbade, Pomrenze, and Silber, 1979;
Glosten and Milgrom, 1985).4
Flood (1994) investigates the intraday operational eciency of the U.S.
foreign-exchange market by executing simulation experiments calibrated on
a particular market structure for market makers, brokers, and customers.
His results support the traditional view that centralization is a key factor
in achieving operational eciency. In particular, he nds that signicant
operational ineciencies may be explained by temporary inventory imbalances inherent in a decentralized market. He also nds that this ineciency
could be largely alleviated by centralizing price information.
Interesting work on the implications of decentralization in this context
has been done by Perraudin and Vitale (1996), who construct an elegant
theoretical framework for a decentralized dealer market and discuss the
implications for price-information-transmission mechanisms and eciency
in the foreign-exchange market. They model interbank trading so as to show
that market makers sell each other information about their transactions
with outside customers by adjusting the bid and ask prices and the bidask spread in their price quotations (for example, by lowering the bid and
ask prices for a particular currency if the bank has been given a large
amount of that currency). This modeling assumption in turn implies that
market makers can capture, through interdealer trades, the informational
rents associated with receiving outside orders and that they can optimally
adjust bid-ask spreads to maximize those rents and elicit information. The
model also implies that bid-ask spreads are wider in decentralized markets,
because, by using wider spreads, dealers can discourage liquidity traders,
who are strongly reactive to price changes, and can therefore increase the
informativeness of their order ow.5 Perhaps the most important nding
from Perraudin and Vitale, however, is that decentralized markets are much
less subject to crashes than are centralized markets. The intuition behind
this conclusion is that information on order ow is used to update subjective
4 One of the benchmark models in the literature is Glostens and Milgroms (1985)
sequential-trade model, which is designed to capture the way in which bid and ask prices
evolve in a dealership market as orders arrive in sequence. For a comprehensive discussion
of the Glosten-Milgrom model and its subsequent extensions, see Lyons (2001).
5 Liquidity traders can be thought of as agents who need to buy or sell foreign exchange
in response to international trade in goods and services and who do not speculate or hedge
their exposure in any way. Their orders are therefore largely random.

estimates of the underlying value of exchange rates. Thus, even when a

centralized market can be expected to collapse from the excessive number
of informed traders, dealers in a decentralized market will still have an
incentive to preserve some turnover, because they can use the information
in the order ow in future trading. Finally, the Perraudin-Vitale model also
suggests a dierence in exchange-rate behavior between the two types of
market system. In a centralized market, bid and ask quotes are martingales
with respect to the information available to foreign-exchange dealers; in a
decentralized market, however, bid and ask spreads shrink as order ow
provides new information.
One important consequence of decentralization in the foreign-exchange
market is a degree of fragmentation; because not all dealer quotes are observable, transactions may occur at the same time at dierent prices. This
feature strongly distinguishes the foreign-exchange market from other nancial (for example, equity) markets. Another important implication of
decentralization in the foreign-exchange market is that this market has less
transparency than other nancial markets, in which trades must, by law,
be disclosed within minutes. Because foreign-exchange markets have no
disclosure requirement, trades are not observable. As a consequence, order
ow cannot communicate information about fundamentals.
Another important feature of the foreign-exchange market is that market makers, who by denition provide two-way bid and ask prices, largely
dominate the market at commercial and investment banks. Market makers
account for some 90 percent of overall trading volume, either by trading
with each other directlymore than 50 percent of the spot-market trading volumeor through foreign-exchange brokersabout 40 percent (BIS,
1998). In the direct market, banks approach each other, and the bank receiving a call acts as a market maker for the currency to be traded by
providing bid and ask prices for the bank placing the call. Brokered transactions occur when brokers collect limit orders from banks. The limit orders, specifying the quantity and the price of an oer to buy or sell, remain
with the broker until they are withdrawn by the bank.6 Brokers trade
by keeping a book of banks limit orders for buying and selling from
which they quote the inside spread (that is, the best bid and ask orders)
on demand. Traditionally, the brokered market has been conducted by
voice through direct telephone links between brokers and banks, although,
as we discuss below, an important recent feature of the foreign-exchange
market has been the development of electronic screen-based brokering sys6 The advantages of brokered transactions are well known and include the rapid dissemination of orders to other market makers, anonymity in quoting, and freedom not to
quote to other market makers on a reciprocal basis, a reciprocity that may be required
in direct transactions.

tems. Only a small fraction of spot-market volume (about 5 percent in the

United States, for example) is traded by bank customers for the purpose of
completing transactions in international trade (goods-market transactions).
Finally, the other participants in the foreign-exchange market are central
banks that trade mainly to complete their own transactions or to inuence
exchange-rate movements through ocial intervention.7
Another classifying distinction is between single-auction and doubleauction systems. In the single-auction system, prices are specied either to
buy or to sell, but not both. In the double-auction system, market makers
provide two-way prices on both bid and ask. Clearly, in the direct foreignexchange market, market makers provide double-auction prices, whereas
the brokered market is largely single-auction.8
With regard to communicating prices, nancial markets are generally
classied as having either open or closed (or public or condential)
order-book systems. In an open system, oers to buy or sell at a specied
price are announced to all agents in the market. In a closed system, offers will be known only to the entity placing the order and perhaps to a
disinterested auctioneer. Clearly, the direct market is an open-order-book
market, and the brokered market is a closed-order-book market. In principle, however, the optimal system of price communication is the one that
minimizes the costs arising from searching out a counterparty in a transaction. Given that the foreign-exchange market is not fully centralized, it
might be argued that the system is not cost-minimizing.9
7 Although market makers may trade for themselves, brokers do not, but prot, instead, by charging a fee for bringing market makers together. Another distinction made
in the microstructural nance literature, originally by Garman (1976) and Schwartz
(1988), is between a continuous (asynchronous) market and a call (synchronous) market
in order to categorize markets. In a continuous market such as the foreign-exchange
market, trade is allowed to occur at its own pace, and transactions are processed as they
arrive. In a call market, trade occurs only at predetermined times, and transactions are
detained until the next call.
8 Note that the distinction between single-auction and double-auction may not be as
clear cut as it rst seems, because, in principle, there is nothing to prevent a bank from
acting as a market maker by simultaneously placing both bid and ask limit orders. The
taxonomy remains useful, however, if we see the distinction between a double-auction
and a single-auction market as being that the former requires a two-way price to be
made, whereas the latter does not (although it may not preclude it). The type of
auction used in a market is also closely related to the degree of centralization of the
market itself. As noted by Flood (1991), for example, the absence of market makers in a
single-auction market and, therefore, the presence of larger search costs lead to greater
centralization of price information. Decentralization of price information, however, leads
to a tendency toward double-auction prices in order to facilitate the searching process
for the counterparty (Goldberg and Tenorio, 1997, and the references therein).
9 This distinction may become obsolete in the future, however, and readers should be
aware that the use of these terms may be shifting (Glosten, 1994). The distinction is
made here as a classicatory device that is useful for the present survey.

One issue that seems not to have been fully examined in this connection
is the importance of credit and the role of clearing houses.10 In a decentralized market, where trade occurs in private meetings between parties, an
assessment of credit risk (whether the counterparty will deliver the agreed
amount of funds at the agreed time at the agreed location) is presumably
made before a deal is struck. In fact, a bank will usually have a prudential
limit, determined by its senior management, on the credit exposure it is
willing to have with any particular counterparty at any point in time, and
it will not deal with that counterparty once this limit is reached until the
net exposure is diminished.
The existence of such prudential limits may be problematic for a fully
centralized system, in which prices are public information and the counterparty is known only after the deal is struck. In practice, centralized
nancial exchanges often circumvent this problem by having a clearinghouse system. For example, the London Clearing House (LCH) acts as
the central counterparty for trades conducted on several of the London nancial exchanges, including the London International Financial Futures
and Options Exchange. This enables the LCH to guarantee the nancial
performance of every contract registered with it by its members (the clearing members of the exchanges) up to and including delivery, exercise, and
settlement. Both sides of a trade are entered in an electronic registration
system by the exchange members, the exchange itself, or both, where they
are matched and conrmed. At the end of each business day, the trade
is registered in the name of the LCH, and the original contract between
the two agents is canceled and replaced by two new onesone between the
LCH and the clearing member seller, the other between the LCH and the
clearing member buyer. This is known in legal terms as novation.
A clearing-house system thus serves two functions in a centralized market. First, it obviates the need to assess the credit rating of the counterparty
or, indeed the need to identify the counterparty at allbefore a trade
is made. Second, because all trades are eectively made with the clearing
house, only the net position at the end of the day is important for settlement purposes. In a decentralized system, trades may be made with a
great many counterparties, so that, even though the end-of-day net position may be close to zero, the trader may have lled a number of credit
limits and will still have to complete all of the agreed-upon trades (with the
consequent transactions costs). In the brokered foreign-exchange market,
although dealers trading on a price quoted by a broker will not know the
identity of the counterparty until after the trade has been made, trades
occasionally have to be canceled once the identities are revealed and one
1 0 We

are grateful to an anonymous referee for encouraging us to pursue this issue.

party nds that its credit limit with the counterparty is full.11 The need
to assess creditworthiness prior to trading may thus explain why the bulk
of foreign-exchange trading (63 percent in April 1998) is between dealers
(BIS, 1998). Moreover, in the absence of an international clearing-house
system, market makers in the foreign-exchange market may fulll many of
the functions of a clearing house. This may explain why the market allows
them to earn signicant prots on the bid-ask spread. The issue of credit
and creditworthiness may thus explain many of the institutional features
of the foreign-exchange market, including the predominant system of communication of prices (open-order-book and interdealer) and the importance
of established market makers. This is clearly an issue that would repay
further research.
Another important feature of the foreign-exchange market is the development, since 1992, of proprietary screen-based electronic foreign-exchangedealing and brokering systems such as the Reuters 2000-2 Dealing System
(Reuters) and the Electronic Broking System Spot Dealing System (EBS).
These systems may also be seen to perform many of the functions of a
clearing house and are an important driving force toward the virtual centralization of the global foreign-exchange market. The Reuters and EBS
are anonymous interactive brokering and dealing systems delivered over
proprietary networks and combining standard closed-order-book brokering
services with the opportunities for electronic trading. Deals are completed
by keystroke or automatic deal-matching within the system. An important
feature of these systems is their prescreened credit facility, so that although
the closed-order-book prices on the dealers screen seem to be anonymous,
the system has automatically checked that the bank to which the price is
being fed has a sucient open credit limit with the originator of the price.
This feature eliminates the potential for failed trades by assuring traders
that they can deal on the prices they see. Credit limits are set and maintained by the designated bank sta and can be modied at any time during
the trading day. Access to the credit le is password driven and condential.
An automated electronic interface between the system and a banks internal banking system, moreover, allows for net, rather than gross, transaction
processing between counterparties and eliminates duplicate trade-entry activity. Although this is not quite the same as dealing with a clearing house,
it substantially reduces transactions costs. These electronic dealing and
brokering systems appear to be growing rapidly in use. Reuters currently
oers its 2000-2 service in forty countries, and the EBS has 800 subscriber
1 1 In practice, where a limit order is placed with a broker by a small or exotic bank,
the broker may try to circumvent this problem by quoting the inside spread, ignoring
the limit order, adding that a better bid or oer is available for a certain amount from
an exotic, and allowing dealers to ask the identity of the potential counterparty before
the trade is made.

banks throughout the world, using some 2,500 workstations to transact

average daily volumes in excess of $90 billion (although this is still very
small in comparison to the total turnover of daily foreign-exchange transactions).12 It seems likely that the continued development of these services
may lead to the virtual centralization of the foreign-exchange market
through the computer network.
Overall, the direct foreign-exchange market may be classied as decentralized, open order-book, and double-auction, whereas the brokered
foreign-exchange market may be classied as quasi-centralized, closed-orderbook, and single-auction. The growing importance of electronic brokering
services suggests that the foreign-exchange market may, in the future, become increasingly centralized.
Foreign-Exchange-Market Activity
The Bank for International Settlements (BIS) regularly collects and documents data on exchange-traded business from individual foreign-exchange
markets. This section provides a brief overview of foreign-exchange-market
activity largely based on the banks Central Bank Survey of Foreign Exchange and Derivatives Market Activity (1998).
Global turnover in traditional foreign-exchange-market segments (spot
transactions, outright forwards, and foreign-exchange swaps)13 was, in 1998,
about $1.5 trillion per day. Forward instruments (outright forwards and
foreign-exchange swaps) were in a dominant position relative to spot transactions, with a market share of 60 percent (which seems still to be increasing), and the foreign-exchange market was dominated by interdealer
business (about 63 percent) and cross-border transactions (about 54 percent). The falling spot-market share was caused, not by a reduction in
turnover in absolute terms, but by the rapid increase in the turnover of
derivatives markets.14
1 2 These gures were taken from the EBS and Reuters websites, at www.ebsp.com and
www.reuters.com/transactions/d22s.htm, respectively. Here and throughout, billion
equals one thousand million.
1 3 Spot transactions are outright transactions involving the exchange of two currencies
at a rate agreed to on the date of the contract for value or delivery (cash settlement)
within two business days. Outright forwards are transactions for value or delivery at
some time in the future (more than two business days later). Foreign-exchange swaps
are the exchange of two currencies (principal amount only) on a specic date at a rate
agreed to at the conclusion of the contract (the short leg), and a reverse exchange of
the same two currencies at a date further in the future at a (generally dierent) rate
agreed to at the time of the contract (the long leg). Foreign-exchange swaps are now
quite common, and the two parties involved are typically rms in dierent countries,
each of which nds it easier (less costly) to borrow in its home country. If, however, each
party wishes to borrow the foreign currency at the foreign interest rate, then a swap may
create gains for both parties (Dubofsky, 1992, pp. 652-656).
1 4 Tight arbitrage relations link the foreign-exchange markets submarkets, which dier

The U.S. dollar was, as one would expect, the most actively traded
currency in 1998, being involved in 87 percent of all transactions worldwide.
The deutsche mark followed, with almost 33 percent of all currency trades
(although the mark was also increasingly used as a proxy for the euro in
the run-up to the euros launch in January 1999). The yen was the third
most used currency, accounting for about 21 percent of all currency trades.
The British pound was the fourth, with 11 percent of trades.
Foreign-exchange-market turnover showed an increasing concentration
in four centers: the United Kingdom (32 percent), the United States (18
percent), Japan (8 percent), and Singapore (7 percent). Although some
smaller countries experienced low or negative growth during the 1990s (for
example, Austria, Finland, and Sweden), a number of others realized quite
spectacular gains, especially in the second half of the 1990s (for example,
Greece, Ireland, the Netherlands, Portugal, and South Africa).15
The replacement of a number of European currencies by the euro, on
January 1, 1999, has considerably modied the conguration of markets
and related trading strategies, as well as the evolution of risk exposure and
management. Indeed, information about the relative importance of the
euro as a traded currency during its early years of operation will be one of
the most widely awaited features of the 2001 BIS survey.
Overall, however, the information available suggests that foreign-exchange-market activity is continuing to grow, largely in response to the
rapid growth of derivatives products. Understanding the future evolution
of the foreign-exchange market, as well as the eects of the emerging euro
market, poses new challenges as well as new opportunities for both practitioners and academics.

in microstructure. Although the literature covered in this study has focused largely on
the spot market, extending the analysis to derivatives markets would open an important
avenue of research. Our interpretation of volume shares from dierent instruments is
essentially literal. It is arguable, however, that swaps, for example, do not involve net
demand in one direction or the other. Rather, a currency swap can be replicated by
buying a bond in one currency and selling a bond in another; the arbitrage does not
involve a net foreign-exchange transaction. Currency swaps, therefore, with their large
notional values, might lead us to underestimate the importance of the spot market in
terms of net order ow (that is, in terms of market-clearing net demand). We are grateful
to Richard Lyons for pointing this out.
1 5 Although the advent of the euro may appear to have caused the decline in some
European centers, the strong growth rates experienced by Ireland, Portugal, and the
Netherlands suggest that other factors are in play. Portugals growth, in particular, might
be attributed to the positive eects of the liberalization of the foreign-exchange market
in 1992. The Netherlands also saw strong growth in its local market as institutional
investors expanded their international portfolios.

10

3 EXCHANGE-RATE EXPECTATIONS
Survey-Data Studies
The process by which agents form expectations about future exchange rates
is the subject of much of the early literature on foreign-exchange-market
microstructure. The simplifying assumption that underlies virtually all
the traditional asset-based exchange-rate models is that expectations are
rational. Thus, the expected exchange rate has usually been measured
in empirical studies using either the forward discount or the interest-rate
dierential, with uncovered interest-rate parity or uncovered and covered
interest-rate parity being implicitly assumed (Taylor, 1995). The strategy
followed by much of the microstructure literature has been to employ direct
measures of expectationsthat is, data on exchange-rate expectations from
surveys of market participants conducted by nancial-services companies
(see Takagi, 1991, or Frankel and Rose, 1995, for an overview of surveydata studies).
Although the various studies using these data dier in their specic
strategies and results, some general qualitative results are discernible. All
foreign-exchange-survey data sets, for example, suggest a strong heterogeneity of expectations and an increasing dispersion of expectations at
longer forecast horizons. They also show a twist in expectations, in
that long-run expectations tend to reverse the direction of short-run expectations. In other words, although a depreciation is usually followed by
expectations of further depreciations in the short run, it is usually followed
by expectations of a moderate appreciation in the long run.1
Survey data have mainly been used to study the existence of foreignexchange-market risk premia in spot-forward regressions and the validity of
the rational-expectations hypothesis (Taylor, 1995). The main conclusions
of these studies are that (1) survey data generally suggest the presence of
1 The discrepancy between the mechanisms for the short- and long-run exchange-rate
expectations formation used by agents in the foreign-exchange market suggests that
short-run expectations generally respond to lagged exchange-rate changes by moving in
the same direction (away from the long-run equilibrium exchange rate), whereas longrun exchange-rate expectations respond to lagged exchange-rate changes by moving in
the opposite direction (toward the long-run equilibrium exchange rate). This behavioral
discrepancy has been rationalized by Froot and Ito (1989) using the concept of consistency, which may be considered a weaker condition than full rationality. Nevertheless,
Froot and Ito (1989, p. 506) nd that expectations generally do not satisfy the consistency condition: In every one of twenty sets of time-series estimates encompassing four
surveys, ve forecast horizons and ve currencies, shorter-term expectations overreact
relative to longer-term expectations when the exchange rate changes. See also Pesaran
(1989) for an alternative derivation of the consistency condition.

11

a nonzero risk premium, which appears to be stable and uncorrelated with

the forward discount, and (2) tests of the rational-expectations hypothesis
generally reject both the unbiasedness condition (the condition that the
expected spot exchange rate is an unbiased predictor of the future spot exchange rate) and the orthogonality condition (the condition that prediction
errors are uncorrelated with any variable in the set of available current information at the time the prediction is made; see Takagi, 1991, and Taylor,
1995). These studies also agree in strongly rejecting both static expectations, where the expected change is always zero, and bandwagon eects,
where the expected change is always greater than the most recent change
(for example, Frankel and Froot, 1986, 1987, 1990a, 1990b; Dominguez,
1987; Allen and Taylor, 1990; Ito, 1990; Taylor and Allen, 1992).
Overall, this strand of the literature provides clear evidence for the existence of risk premia and against the pure rational-expectations hypothesis.
In addition, it suggests that the true, unknown, expectations-formation
process used by agents in the foreign-exchange market is likely to be more
complex than other standard alternatives, and that heterogeneity of expectations may be crucial (Taylor and Allen, 1992). Although additional
empirical work on survey data is needed to establish the robustness of these
ndings, a theoretical inquiry into the formation of expectations and the
causes and implications of their heterogeneity would seem to be an immediate avenue for future research.
Order-Flow Studies
A related literature has shown that time-aggregated order-ow variables
may be more powerful than macroeconomic variables in explaining exchangerate behavior. Order ow, which in this context is taken to be a variant
of the more familiar concept of net demand, measures the net of buyerinitiated and seller-initiated orders. As noted by Lyons (2001), it is a variant
of, rather than a synonym for, net demand, because, in equilibrium, order
ow does not necessarily equal zero.
The fact that order ow helps explain exchange-rate behavior does not
necessarily imply that order ow drives exchange rates. Indeed, it may
well be that macroeconomic fundamentals are the driving force, but that
conventional measures of the macroeconomic fundamentals are so imprecise
that an order-ow proxy performs better in estimation. The proxy interpretation is particularly plausible with respect to expectations, for even
if macroeconomic variables fully describe the true model, they may be a
poor measure of expected future fundamentals when implemented empirically. Unlike expectations measured from survey data, moreover, order
ow represents a willingness to back ones beliefs with real money (Lyons,
2001).
12

A model recently proposed by Evans and Lyons (1999) presents order

ow as a proximate determinant of price. Using data on signed order ow
from the Reuters dealing system, Evans and Lyons show that order ow is
a signicant determinant of some leading bilateral exchange rates, and that
the coecients of determination are signicantly larger than those usually
obtained using standard macroeconomic models of nominal exchange rates.
The authors also show that their model of daily exchange-rate changes
produces good out-of-sample forecasts at short horizons, beating the alternative of a random-walk model in standard measures of forecast accuracy.

13

4 CHARTIST ANALYSIS
An alternative explanation of the discrepancy between short- and longrun exchange-rate expectations is that the participants in foreign-exchange
markets may use dierent forecasting techniques for dierent expectations
horizons. The predominant technique used for short-run exchange-rate forecasting may be chartist analysis, for example, whereas the technique used
for long-run exchange-rate forecasting may be based on fundamental analysis or on conventional exchange-rate models.
Chartist, or technical, analysis uses charts of nancial-asset price
movementsoften with the aid of additional descriptive statisticsto infer
the likely course of future prices and thus to derive forecasts and trading
strategies. The trends and patterns that chartists consider are generally
inferred through loose inductive reasoning. Perhaps the most famous example of a chartist technique (indicating a trend reversal) is the head and
shoulders pattern presented by Allen and Taylor, (1990; Taylor and Allen,
1992). Chartists often use trends and patterns to identify broad ranges
within which exchange rates or asset prices are expected to trade, attempting to set the parameters for price movements. They may also employ
one or more mechanical indicators when forming a general view, which may
be trend-following (for example, based on moving averages) or non-trendfollowing (for example, rate-of-change indicators or oscillators used on
the assumption that the markets tend to correct when an asset has been
overbought or oversold). In practice, chartist analysis includes both
pattern and trend recognition, as well as information from basic statistical
indicators.
Chartist Analysis in the Foreign-Exchange Market
Although casual observation suggests that the use of chartist analysis is
widespread in leading nancial markets, evidence about its use has been
largely anecdotal (Malkiel, 1996). A survey conducted by the Group of
Thirty (1985), however, reported that 97 percent of banks and 87 percent of
securities houses believed that the use of technical analysis has a signicant
impact on the foreign-exchange market.
Applied work on the use of technical analysis includes that by Allen
and Taylor (1990; Taylor and Allen, 1992), who carried out a study of the
London foreign-exchange market on behalf of the Bank of England. Their
questionnaire, sent to over 400 chief foreign-exchange dealers in the London market in November 1988, achieved a response rate of over 60 percent
and suggested a broad consensus regarding the weight given to chartist
14

analysis at dierent time horizons. At short time horizons (intraday to

one week), approximately 90 percent of respondents reported using some
chartist input when forming their exchange-rate expectations, with 60 percent judging charts to be at least as important as fundamentals. At longer
forecast horizons, from one month to one year, the weight given to economic
fundamentals increased. At the longest forecast horizons, of one year or
longer, the skew toward fundamentals was most pronounced, with about
33 percent of the respondents relying on pure fundamentals and about 85
percent judging fundamentals to be more important than charts. There
was, moreover, a persistent 2 percent or so of respondents who apparently
never used fundamental economic analysis at any horizon. Other ndings
of the Taylor-Allen study were that dealers perceived chartist analysis and
fundamental (economic) analysis to be complementary approaches and that
signicant numbers of them viewed technical analysis as self-fullling.1
More recent evidence is presented by Menkho (1997, 1998), who examines some of the basic assumptions of the noise-trading approach by means
of a questionnaire sent to foreign-exchange-market participants in Germany.
Menkho identies two groups characterized by their use of dierent information: rational arbitrageurs (who rely primarily on fundamental analysis)
and not fully rational noise traders (who prefer other forms of analysis to
fundamental analysis). The empirical results also imply the existence of
short horizons and sentiments assumed by the noise-trading approach, although these cannot be exclusively related to the respective groups. This
nding suggests that rational agents may use nonfundamental analysis to
exploit less-rational noise traders.
Cheung and Wong (1999, 2000) report ndings from a survey of practitioners in the interbank foreign-exchange markets in Hong Kong, Tokyo,
and Singapore. The majority of respondents agree that nonfundamental
factors have signicant eects on short-run exchange rates and that speculation increases volatility while improving liquidity and eciency.2 Another
study, by Cheung and Chinn (1999), presents ndings from a survey of practitioners in the U.S. interbank foreign-exchange market. The results from
1 In addition, Allen and Taylor (1990) analyze the accuracy of individual technical
analysts one-week-ahead and four-week-ahead forecasts of three major exchange rates.
They nd that, over a ten-month period, some individuals were able to outperform a
range of alternative forecasting procedures, including a random-walk model, vector autoregressions, and univariate autoregressive moving-average time series. See also Curcio
and Goodhart (1992, 1993).
2 In a related study, Lui and Mole (1998) report the results of a survey conducted in
February 1995 on the use by foreign-exchange dealers in Hong Kong of fundamental and
technical analysis to form their forecasts of exchange-rate movements. They nd that
about 85 percent of respondents rely on both fundamental and technical analysis, and
that there is a skew toward technical analysis at shorter horizons, and a skew toward
fundamental analysis at longer horizons.

15

this study suggest that only about 30 percent of traders can be characterized as technical traders, although this proportion has been increasing
over the last ve years. Incorporation of news about fundamentals into
exchange-rate expectations occurs very rapidly, and news about interest
rates appears to be relatively more important in these markets. As one
would expect, macroeconomic fundamentals appear to be particularly important for long-run exchange-rate expectations.3
At the very least, therefore, the empirical evidence suggests that the
attitude of many nancial economists toward chartist analysisthat technical strategies are usually amusing, often comforting, but of no real value
(Malkiel, 1996, p. 154)should not be held with 100 percent condence,
at least in the foreign-exchange market.
The Impact of Chartist Analysis
The widespread academic skepticism of nonfundamental inuences in nancial markets has begun to give way to active investigation of the many
phenomena not captured by traditional economic models. Although analyses of the stock market have begun to consider the inuence of nonfundamental factors in general (Shiller, 1984; De Long, Shleifer, Summers, and
Waldmann, 1989), studies of the specic role of chartism have so far been
largely conned to the foreign-exchange market.
Goodhart (1988) discusses the way in which exchange-rate misalignment might occur by considering that the rate might be determined by
a balance of chartist and fundamentalist predictions. This surmise is, in
some ways, similar to the simple model of the stock market suggested by
Shiller (1984), in which the equilibrium price depends on the balance between fundamentalists (smart money) and ordinary investors who subscribe
to popular models. A comparable approach is developed more formally by
Frankel and Froot (1990b), who explain the sharp rise in the demand for
the U.S. dollar over the 1981-85 period as a shift in the weight of market
opinion away from fundamentalists and toward chartists. This shift is modeled as a Bayesian response to the inferior forecasting performance of the
economic fundamentalists.
3 Cheung and Chinn (1999) also report that speculation and ocial intervention, although thought to increase market volatility, are likely to restore equilibrium by moving
exchange rates toward their long-run values. Finally, traders do not value purchasingpower parity highly as a long-run equilibrium concept, although they seem to believe
that it might be useful for horizons of at least six months. Cheung, Chinn, and Marsh
(1999) use U.K. survey data and conrm that nonfundamental factors are thought to
dominate short-run movements in exchange rates, although fundamentals are deemed
important over much shorter horizons than is suggested by the mainstream empirical
literature.

16

Bilson (1990) emphasizes that technical traders employing overbought

or oversold indicators (oscillators) will impart nonlinearity into exchangerate movements, because small exchange-rate changes that do not trigger
the oscillator will tend to be positively correlated, reecting trend-following
trading programs; larger exchange-rate movements, however, which trigger
an oscillator, will be negatively correlated. Bilson estimates simple nonlinear exchange-rate equations that are consistent with this pattern of serial correlation and records moderate success in capturing exchange-rate
changes.4
Researchers have also begun to analyze related microstructure issues
such as the volume of trade in foreign-exchange markets, and to ask why,
for example, trade volume is much higher on a gross basis (among foreignexchange dealers and brokers) than on a net basis (involving nonnancial
companies).5 Frankel and Froot (1990b), for example, provide strong evidence that both volume of trade and market volatility are related to the
heterogeneity of exchange-rate expectations, as reected in the dispersion
in survey expectations (see also Ito, 1990, and Hogan and Melvin, 1994).
Other studies in the microstructure literature have looked at the way
information is processed and transmitted through the market, and at the
eect information processing has on market volatility and volume. Lyons
(1995), for example, reports that volume aects the bid-ask spread through
both the information signaled by market volume and the desire of market
makers to control their inventory of currencies. Bessembinder (1994) reports
that bid-ask spreads widen as proxies for inventory carrying costs increase.
Bollerslev and Melvin (1994) provide a model in which the bid-ask spread is
determined by underlying uncertainty regarding exchange-rate movements.
In related work on market volatility, others document regularities in market
volatility (Hsieh, 1988) and the contagion of exchange-rate volatility across
foreign-exchange markets, the so-called meteor shower, which may be
interpreted as evidence of information processing (Engle, Ito, and Lin, 1990;
Hogan and Melvin, 1994).
Modeling Expectations of Chartists, Fundamentalists, and Portfolio Managers
Frankel and Froot (1987) develop a formal model of agents expectations in
the foreign-exchange marketwith the agents classied as chartists, fundamentalists, and portfolio managersthat forms the basis for nearly all
subsequent models of investors expectations in the foreign-exchange and
4 Hsieh (1988) also detects evidence of nonlinearities in exchange-rate movements. See
also De Grauwe, Dewachter, and Embrechts (1993).
5 Frankel and Froot (1990b), for example, suggest that over 95 percent of market
transactions in leading exchange rates involve foreign-exchange-market dealers or brokers; Lyons (1995) suggests that market makers alone account for over 80 percent.

17

other nancial markets. Their model, designed to explain the demand for
the U.S. dollar during the 1980s, functions as though there are actually
two models of the dollar operating, one at each end of the spectrum, and
a blend in between (Frankel and Froot, 1986, p. 36). The two models
they use are a fundamentalist model, for which they assume a Dornbusch
overshooting model (long-run), and a chartist model, for which they use
a simple autoregressive integrated moving average (ARIMA) forecasting
equation (short-run).
The value of a currency can then be driven by the decisions of portfolio
managers who consider a weighted average of the expectations of fundamentalists and chartists of the following form:
f
c
sm
t+1 = t st+1 + (1 t ) st+1 ,

(1)

f
c
where sm
t+1 , st+1 and st+1 denote the rate of change in the spot rate expected by portfolio managers (the market), fundamentalists, and chartists,
respectively, and t is the weight given to the fundamentalist view by portfolio managers. Assuming, for simplicity, that sct+1 = 0, equation (1)
becomes
f
(2)
sm
t+1 = t st+1 ,

and thus

f
t = sm
t+1 /st+1 .

(3)

Consider a general model of exchange-rate determination of the form

st = csm
t+1 + zt ,

(4)

where zt denotes the fundamentals. Clearly, equation (4) is a very general

exchange-rate model that can be interpreted in terms of any of the assetpricing exchange-rate determination models. The Frankel-Froot model incorporates an expectations-formation mechanism of the form (1) within a
general asset-pricing model of the type (4).
Frankel and Froot (1987) analyze the simple case in which chartists
believe that the exchange rate follows a random walk, sct+1 = 0, assuming
a process governing the evolution of the weight t and using simulation
methods, rather than computing the closed-form solution of the model.
They also fully specify a mechanism to endogenize the fundamentals in
order to include in the model the eects of current-account imbalances,
allowing the currency to revert to equilibrium and avoiding the drawbacks
of an exogenous specication of the economic fundamentalsin particular,
the implication that the spot rate may be stuck at a disequilibrium level.6
6 Note also that the model assumes behavior that is less than fully rational, because
none of the three classes of agents conditions its forecasts on the full information set of
the model.

18

The results of the simulations suggest that a bubble in the exchange rate
is generated by portfolio managers attempts to learn the model. When the
bubble takes o, as well as when it collapses, portfolio managers learn the
model more slowly than when they are changing it by revising the weight
given to fundamentalist and chartist views. When the weight given to the
fundamentalist view leans toward any of its extreme values (zero and unity),
however, the portfolio managers exchange-rate-determination model leans
toward the correct model. In addition, the revisions of the weight become
smaller until the portfolio managers model corresponds perfectly to the
correct one, implying that portfolio managers change the model more slowly
than they learn it.
Following the seminal article by Frankel and Froot (1987), a number
of researchers constructed models to illustrate the role of nonfundamentalist traders in generating bubbles in the foreign-exchange and other nancial markets. These models are generally alike in considering at least two
types of traders who dier in their forecasts or beliefs, and in predicting
an outcome that shows the nonstationary dynamics typical of many nancial markets. Among them, those by De Long et al. (1989, 1990a, 1990b),
Kirman (1991), and Goldberg and Frydman (1993) are worth citing. The
models predict that periods of steady evolution are interrupted by bubbles
and crashes, with noise around the turning point when one regime shifts
to another. Opinions in the market are usually modied endogenously as
a result of the interactions among agents. Agents make mistakes at the
turning points when the opinion of the majority is not clear, but these
episodes are not long enough for learning or for the generation of protable
arbitrage opportunities (Kirman, 1991). De Grauwe and Dewachter (1993)
use a combined chartist-fundamentalist model to show the way in which
chaotic behavior of the exchange rate may be generated; the exchange rate
produced by such a process would be essentially unpredictable, regardless
of whether the underlying model is deterministic or stochastic (Goodhart,
1988). If these models accurately describe the foreign-exchange market, the
discrepancy between short- and long-run expectations of the exchange rate
may no longer be considered a puzzle for the profession.
Summing Up
Although it seems inexorable that economic fundamentals will win in
the long term, it is likely that short-term price movements may be dominated by popular models and theories, one of which is chartist analysis. Clearly, simple reliance on the extrapolation and inductive reasoning
found in chartist analysis is ultimately unsatisfactory. It may be, however,
that chartists (and foreign-exchange dealers in general), by working with
the minutiae of market-price movements, are able to get a feel for lo19

cal approximations to processes that are too complex, too short-term, or

too nonlinear to be captured adequately by the current state of nancial
economics. In any event, the fact that chartist techniques are subscribed
to by large numbers of nancial-market practitioners, combined with the
poor performance of standard exchange-rate economics in predicting and
explaining exchange-rate movements, suggests that chartism should not be
lightly dismissed (Frankel and Rose, 1995; Taylor, 1995; Evans and Lyons,
1999).

20

5 TIME-VARYING VOLATILITY, MARKET LOCATION,

AND TRADING VOLUME
Time-Varying Volatility in the Foreign-Exchange Market
The modeling of time-varying volatility in the foreign-exchange market is
the focus of an enormous literature that is largely dominated by models
of autoregressive conditional heteroskedasticity (ARCH) and generalized
autoregressive conditional heteroskedasticity (GARCH). The basic insight
behind ARCH models, rst introduced by Engle (1982), is that the second moment of the distribution of exchange rates is serially correlated.
ARCH models are nonlinear conditionally Gaussian models in which the
conditional variance is a function of the lagged error terms. Engles (1982)
ARCH model may be written as follows:
yt | t1 N (xt , ht )

(5)

ht = f (t1 , t2 , . . . , tq , . . . , )

(6)

t = yt xt ,

(7)

implying that the conditional distribution of yt , given the information set

t1 , is normal, with mean and variance equal to xt (a linear function of,
say, k independent variables xt ) and ht (a function of, say, q lagged error
terms and other exogenous variables), respectively. Equation (6) may be
parameterized as

q
ht = 0 + j=1 j 2tj
j 0 j,
(8)
for an ARCH(q) model. These models can be estimated by maximumlikelihood methods, and a simple Lagrange multiplier (LM) test for the
presence of ARCH eects may be constructed from an auxiliary regression
of the form
2t = 0 + 12t1 + . . . + q2tq ,
(9)
where t denotes the residual from ordinary least squares estimation of (7).
Under the null hypothesis that j 0 j, the LM statistic T R2 (T being
the number of observations and R2 being the coecient of determination
from equation 9) is asymptotically distributed as 2 (q 1).
A simple but important extension of the ARCH model is the GARCH
model developed by Bollerslev (1986). A GARCH(p,q) model may be written as




ht = 0 + qj=1 j 2tj + pj=1 j htj .
(10)
21

Because 2t has the same properties as an autoregressive moving-average

(ARMA[m,p]) process, the appropriate orders of a GARCH process may
be determined following the criteria typical of the Box and Jenkins (1970)
methodology on the autocorrelations and partial autocorrelations of the
squares of a set of residuals.1
A common nding of the literature using high-frequency nancial data
is the strong persistence implied by the estimates of the conditional variance functions of excess returns. This persistence is consistent with the
presence of a near-unit root in the autoregressive polynomial in (10), that
is, 1 + . . . + q + 1 + . . . + p = 1. Engle and Bollerslev (1986) depict this situation as an integrated generalized autoregressive conditional
heteroskedasticity (IGARCH) model. The strong implication of this sort
of model is that current information remains important for forecasting the
conditional variance for the indenite future.
Time-varying volatility models of the kinds described above have been
applied to exchange rates by, among others, Domowitz and Hakkio (1985),
Engle and Bollerslev (1986), Milhoj (1987), Hsieh (1988), McCurdy and
Morgan (1988), and Baillie and Bollerslev (1989); for a survey, see Bollerslev, Chou, and Kroner, 1992, pp. 37-46).
Hsieh (1988) uses daily data for ve exchange rates vis--vis the U.S.
dollar from 1973 to 1983. His main purpose is to investigate the statistical
properties of the data and to discriminate between two competing explanations for the observed heavy tails of the distribution of exchange rates:
namely, that the data are independently drawn from a heavy-tail distribution that remains xed over time and that the data come from distributions
that vary over time. Hsiehs results suggest the rejection of the former hypothesis. An interesting nding is that the rejection can be attributed to
changing means and variances in the data, and that an ARCH(9) model is
able to capture most of the nonlinear stochastic dependencies present in the
data. Similar results are provided by Milhoj (1987), Diebold (1988), and
Diebold and Nerlove (1989). These ndings have been reinforced by other
studies using GARCH formulations, for example, McCurdy and Morgan
(1988), Hsieh (1989a, 1989b), and Kugler and Lenz (1990).
Diebold and Nason (1990) argue, however, that it is not clear whether
the conditional heteroskedasticity detected in the prediction error of linear exchange-rate models is, in fact, a property of the true data-generating
process or whether it is attributable to some sort of general misspecication associated with linear conditional-mean representations. Diebold and
1 For comprehensive textbook expositions of ARCH and GARCH models, as well as
their properties and their extensions, such as ARCH in mean (ARCH-M) and GARCH in
mean (GARCH-M), in which the basic ARCH and GARCH frameworks are extended to
allow the mean to depend on its own conditional variance, see Engle, Lilien, and Robins
(1987), and Bollerslev, Chou, and Kroner (1992).

22

Nason examine this issue by estimating nonparametrically the conditionalmean functions of ten nominal dollar spot rates for the sample period 1973
to 1987, used to generate in-sample and out-of-sample nonparametric forecasts. Interestingly, their ndings do not support the idea that nonlinearities in exchange rates can be exploited for prediction purposes. Their
results are also consistent with the ndings of other similar studies, such
as that by Meese and Rose (1991).
Another nding, suggested by both Diebold (1988) and Baillie and
Bollerslev (1989), is that although ARCH eects are found to be strongly
statistically signicant on daily and weekly exchange-rate data, they weaken
and eventually disappear with less frequently sampled data, that is to say,
ARCH eects aggregate out over time. In addition, the assumption of
normality appears to be a good approximation over four-week or, perhaps,
two-week frequencies, but not at higher frequencies. Baillie and Bollerslev
(1989) also estimate a GARCH(1,1) model and report a value of 1 + 1 very
close to unity, suggesting an IGARCH process (see also Diebold, 1988).2
Andersen and Bollerslev (1998) have recently raised the level of analysis
well beyond the standard GARCH approach. Using an annual sample of
ve-minute returns, their empirical model captures the intraday activity
patterns in the deutsche mark-dollar exchange market, the macroeconomic
announcements in the market, and the volatility persistence (ARCH) known
from daily returns. The authors quantify the dierent features and show
that they account for a substantial fraction of return variability, both at
the intraday and daily level (see also DeGennaro and Shrieves, 1995, and
Melvin and Yin, 1999).
ARCH models of the exchange rate also have important implications
for foreign-exchange-market eciency. The relevant empirical literature
systematically nds that the forward rate is not an unbiased predictor of
the corresponding future spot rate. Under the rational-expectations hypothesis, however, the existence of a risk premium can still reconcile this
fact with eciency in the foreign-exchange market (Hakkio, 1981; Hodrick
and Srivastava, 1984; Domowitz and Hakkio, 1985; Baillie, 1989). The empirical literature has attempted several dierent specications that make
the risk premium a function of the time-varying conditional variance of the
spot exchange rate. In particular, a number of authors have used ARCH2 A nice theoretical contribution in this context has been made by Hodrick (1989),
who studies the failure of log-linear exchange-rate models of the 1970s and the observed
variability of the risk premium in the foreign-exchange market. Rational, maximizing
models predict that changes in conditional variances of monetary policy, government
spending, and income growth aect risk premia, generating conditional volatility of exchange rates. Hodrick examines theoretically the eects these exogenous conditional
variances have on the level of the current exchange rate and quanties the extent to
which this channel explains exchange-rate volatility using ARCH models.

23

style models for the risk premium, with largely unsatisfactory results. Several researchers argue, however, that these weak results may simply be
attributable to the fact that the conditional variances are poor proxies for
risk. In principle, therefore, a risk premium may be modeled more satisfactorily by making it a function of time-varying cross-currency conditional
covariances rather than of just its own conditional variance. Following this
argument, several researchers have estimated multivariate ARCH models
as a test of the foreign-exchange-market eciency hypothesis (Lee, 1988;
Baillie and Bollerslev, 1990). Tests of the conditional capital-asset-pricing
model (CAPM) that allow for a time-varying conditional covariance matrix
have been made, for example, by Mark (1988) and Giovannini and Jorion
(1989) and perform much better than the traditional CAPM (Kaminsky
and Peruga, 1990).
Multivariate ARCH models have also been used to investigate various
policy issues associated with the foreign-exchange market. Examples are
models by Diebold and Pauly (1988) and Bollerslev (1990), who analyze
the short-run volatility of the exchange rate immediately after the inception of the European Monetary System (EMS). Their studies conclude that
an increase after 1979 in the conditional variances and covariances among
the member countries of the EMS occurred as a result of increased policy
coordination.
Volatility and Market Location
What are the implications of time-varying volatility in the foreign-exchange
market? Engle, Ito, and Lin (1990) argue that in an ecient foreignexchange market, the ARCH eects characterizing high-frequency data may
be attributed to the amount, or quality, of information reaching the market,
or else to the time necessary for market agents to process the new information fully. This argument seems consistent with the observation rst
made by Fama (1970, p. 396) about volatility clusters, that large daily
price changes tend to be followed by large daily changes. The signs of the
successive changes are apparently random, however, which indicates that
the phenomenon represents a denial of the random walk model but not of
the market eciency hypothesis. Nevertheless, it is interesting to speculate
why the phenomenon might arise. Baillie and Bollerslev (1991) use fourhourly exchange-rate series for a six-month period in 1986 and develop a
seasonal GARCH model to describe the time-dependent volatility of each
exchange-rate series. Their empirical results suggest that hourly patterns
of volatility are remarkably similar across countries and are strongly associated with the opening and closing of the leading world markets. In
addition, they nd that the U.S. foreign-exchange market displays more
volatility than the European market shows.
24

Engle, Ito, and Lin (1990, p. 526; Ito, Engle, Lin, 1992) use meteorological analogies to explain their viewpoint: news follow[s] a process like
a heat wave so that a hot day in New York is likely to be followed by another hot day in New York but not typically by a hot day in Tokyo. The
alternative analogy is a meteor shower which rains down on the earth as it
turns. A meteor shower in New York will almost surely be followed by one
in Tokyo. To anticipate our conclusion, volatility appears to be a meteor
shower rather a heat wave.3
Engle, Ito, and Lin use intradaily data on the yen-dollar exchange rate
from October 3, 1985, to September 26, 1986, and dene four separate
market locations: Europe, New York, the Pacic, and Tokyo. They then
consider a nonoverlapping market for a day, with market 1 being open
rst. Volatility generated into a previously open market can be considered
exogenous and part of the information set for market 2 on the subsequent
day. The Engle-Ito-Lin model is a GARCH-based vector autoregression
model for per-hour volatility of the form
i,t
hi,t

i,t N (0, hi,t )

i = 1, 2, . . . , n

i1

n
= i + i hi,t1 + j=1 ij 2j,t + j=1 ij 2j,t1 ,

(11)

where j,t is dened as the intradaily exchange-rate change divided by the

square root of business hours in market j on date t; i,t is the information
set for the market segment i at time t, which includes past information until
(t 1) and current information from market 1 to market (i 1), that is,
i,t = {i1,t , i2,t , . . . , 1,t } and is a subset of n,t1 with n > i 1.4 The
heat-wave hypothesis tested by Engle, Ito, and Lin against the alternative
meteor-shower hypothesis amounts to a test of H0 : ij = 0 jointly for
i = j.
Another innovation of the Engle-Ito-Lin study is that they develop a
technique for analyzing the dynamic interreaction generated by countryspecic news. The intuition is to assume that the expected future per-hour
variance in a market segment is a function of a shock to a dierent market
3 Similar arguments are advanced in the theoretical models of foreign-exchange-market
dynamics built by Kyle (1985) and Admati and Pfeiderer (1988). These models have the
peculiar feature that prices do not fully incorporate all the information available about
the fundamental value until the end of trading, a lag that explains volatility spillovers.
These models are therefore consistent with semi-strong-form eciency of the foreignexchange market, but not with strong-form eciency.
4 Various assumptions underlie the model. In particular, the authors assume market
eciency, which implies that intraday exchange-rate changes are distributed with mean
zero and that
i,k and
i,k are uncorrelated for i
= j. Engle, Ito and, Lin also derive
the likelihood function and a maximum-likelihood estimator, or MLE (which is simply
an extension of the MLE developed by Engle and Bollerslev, 1986), in order to estimate
their model.

25

segment. Thus, using vector notation, equation (11) may be rewritten as

ht+1 = + Bht + At+1 + Ct ,

(12)




where is a vector of constants, ht = (h1,t , . . . , hn,t )
, t = 21,t , . . . , 2n,t ,
and

0
0
0
... 0
11 0
0 ... 0
21 0
0

0
... 0
22 0 . . . 0

A = .
,
B
=

,
..
..
..
..

.
.
.

n1

n2

11
0
..
.

12
22

...

... ... 0

. . . 1n
. . . 2n

.
..

. . . nn

(13)

. . . nn


Dening ht+s/nt E ht+s | n,t and taking the iterated expectation of
ht+s , Engle, Ito, and Lin prove that the vector of conditional heteroskedasticity of all the markets follows a process of the form
(I A) ht+s/kt = + (B + C) ht+s1/kt .

(14)

Thus, if Rik (s) hi,t+s/kt /2k,t (for i, k = 1, . . . , n) is the impulse response function of per-hour volatility of market i to the squared innovation
of market k, taking the derivative of (14) allows us to obtain Rik (s) by
solving recursively the equation
(I A) Rk (s) = (B + C) Rk (s 1)

s 2,

(15)

where Rk (s) = [Rik (s)]nx1 . Using simulation methods, Engle, Ito, and
Lin could derive per-hour volatility in each market segment in response
to per-hour volatility of the other market segments and calculate impulse
responses.
Applying their model to yen-dollar intradaily exchange-rate data, the
authors strongly reject the heat-wave hypothesis and interpret this rejection
as consistent either with market dynamics exhibiting volatility persistence
(caused, for example, by private information or heterogenous beliefs) or
with stochastic policy coordination or competition. Thus, they investigate
the dynamic eect of country-specic innovations on conditional volatility
in the subsequent markets. Their ndings suggest that Tokyo news has
the largest impact on volatility spillovers of the yen-dollar exchange rate.
Finally, they compute the impulse response curves to examine the reaction

26

of one markets volatility to news coming from another market. The empirical results suggest a cross-country dynamic eect in the short run, which
gradually dies out. Overall, therefore, the authors nd a case for volatility
clustering of the meteor-shower type, rather than of the heat-wave type.5
Public Information or Private Information?
A common feature of empirical and theoretical studies is the crucial role of
information, especially macroeconomic news, in determining price volatility (Oldeld and Rogalski, 1980; French and Roll, 1986; Harris, 1986).
It is widely thought (and implicitly assumed in conventional macroeconomic models of exchange-rate determination) that all agents in the foreignexchange market base projections on the same public-information set, implying that private information is irrelevant. Some recent studies, however,
challenge this view.
Harvey and Huang (1991) study the volatility implications of foreignexchange trading, using transaction data on futures contracts from the
Chicago Mercantile Exchange and the London International Financial Futures Exchange. To be precise, they analyze the role of public news announcements in determining volatility patterns. Harvey and Huang compute the hourly variance rates for exchange-trading and non-exchangetrading intervals and compute the ratios of both these and the total variance rates.6 The rationale underlying the computation is the so-called
public-information hypothesis, which suggests that even if trading outside exchange-trading time is important, the exchange-trading variance rate
may exceed the non-exchange-trading variance rate, because individual exchange rates are aected by public information available both to the countries concerned and to other countries as well. The availability of this public
information, released during trading hours in both countries, may thus be
expected to aect the volatility of the exchange rate; volatility may increase
when important macroeconomic news is made available.
This view contrasts with that of French and Roll (1986), who suggest
that trading based on private information induces higher volatility when
the market is open than when the market is closed (see also the theoretical
framework of Admati and Peiderer, 1988, and Foster and Viswanathan,
5 See also the closely related study by Hogan and Melvin (1994), who examine the
role that news and heterogeneous expectations play in the meteor-shower eects. Hogan
and Melvin focus on the U.S. trade-balance news, which is shown to have a signicant
and persistent eect on the exchange rate and its conditional variance. In addition, the
impact of U.S. trade-balance news is not isolated to the U.S. foreign-exchange market.
The degree to which this news aects other market locations appears to be functionally
related to heterogeneous priors.
6 The variance-rate ratios are computed under the assumption that returns are serially
uncorrelated.

27

1990). The empirical results of the Harvey-Huang (1991) study provide

strong evidence that the intraday volatility on the international monetary
market varies largely by day of the week, and that the opening on Friday
(to a lesser extent Thursday), in particular, is characterized by very high
volatilities. Harvey and Huang interpret these results as supportive of the
public-information hypothesis, because in the United States, most of the
important public macroeconomic announcements occur on Friday and, to
a smaller extent, on Thursday. Overall, U.S.-European and U.S.-Japanese
exchange-rate volatilities are found to be higher during U.S. trading hours,
and European cross-rate volatilities are found to be higher during European
trading hours. Although the disclosure of private information through trading may partly explain these volatility patterns, macroeconomic announcements appear more likely to be the cause of increases in volatility.
Wasserfallen and Zimmerman (1985) and Goodhart and Giugale (1993)
investigate the systematic patterns of intraday volatility of exchange rates
and conclude that volatility is smaller during intervals when trading volume is known to be smaller (for example, weekends and lunch hours) and
is larger during the rst trading hour on Monday for each currency in the
domestic country, regardless of the fact that other markets have opened earlier. This evidence may be considered supportive of the private-information
hypothesis, although both the ecient-markets hypothesis and the hypothesis that news drives exchange-rate volatility are not entirely discarded (see
Goodhart and Giugale, 1993, pp. 18-19).
Demos and Goodhart (1996) note that trading volume declines immediately before the weekend and report a strong correlation between intraday
patterns in exchange-rate volatility and trading activity: high, for example,
both at the opening and at the close of the market, displaying essentially
a U-shaped pattern. It is not clear, however, whether large volumes and
volatilities are caused by ecient processing of fundamentals or by other
factors such as noise trading or bandwagon eects.
A number of studies in the microstructure literature have reported
a strong contemporaneous correlation between trading volume and the
volatility of exchange rates (Cornell, 1981; Grammatikos and Saunders,
1986; Karpo, 1987). This correlation is important for at least three reasons: First, it provides insight into the structure of nancial markets by
relating new information arrival to market prices. Also it has implications
for the design of new futures contracts; a positive relation suggests that a
new futures contract can succeed only when there is sucient price uncertainty with the underlying asset, which cannot be eectively cross-hedged
with other contracts. Finally, the price-volume relation has a direct bearing
on the empirical distribution of speculative prices (Jorion, 1996, p. 22).
The literature generally suggests that the positive correlation between

28

volume and volatility is driven by an unobserved common driving variable.

This so-called mixture-distribution hypothesis (MDH) was rst proposed
by Clark, 1973, and Epps and Epps, 1976. The seminal theoretical paper
in this literature, by Tauchen and Pitts (1983), assumes that the relation
between volume and volatility may be of two types. First, market-price
volatility is inversely related to the number of traders, because market
prices may be considered to be an average of traders reservation prices,
and an increase in the number of traders is an increase in the number of
observations on which the average (market price) is computed. Second,
given a certain number of traders, price variability is proportionately related to trading volume, because higher trading volume is consistent with
higher disagreement among traders. In addition, the link between trading volume and volatility is stronger when new information ows to the
market at a relatively higher rate.7 Tauchen and Pitts (1983) derive the
joint distribution of daily price changes and transactions volumes from a
model of intraday equilibrium price changes and intraday volumes. Traders
change their reservation prices continuously during the day in response to
new information until the market reaches a new equilibrium.
Frankel and Froot (1990a, 1990b) examine the relation between the dispersion of survey forecast, volatility, and trading volume and nd strong evidence that the dispersion parameter (the disagreement component) aects
both volatility and volume; this nding is consistent with the MDH. Jorion
(1995, 1996), however, argues that implied volatilities may be more informative than time-series models, because forecasts of volatility require forecasts
of parameters that are expected to be time-varying and that are typically
treated as constant in the canonical MDH approach. Given that options
have been traded for about fteen years in the leading stock-exchange markets, there is sucient data available to test time-series models using implied standard deviations (ISDs) in the foreign-exchange market. Wei and
Frankel (1991) and Jorion (1995) examine the predictive ability of ISDs
by relating them to future realized volatility. Their results suggest that
ISDs largely outperform time-series models, although they are still biased
predictors.
In order to compute ISDs, some authors use stochastic-volatility models
7 Unfortunately, the trade-information ow is an unobservable variable, implying that
the estimation of this model requires the assumption of a distribution about an unobserved variable. An alternative model, however, which leads to similar conclusions about
the relation between volume and volatility, is provided by Richardson and Smith (1994;
see also Bessembinder, 1994). Hartmann (1999) analyzes the dollar-yen bid-ask spread,
using a long time series of daily spot foreign-exchange-trading volumes. In line with standard spread models and volume theories, he shows that unpredictable foreign-exchange
turnover (a measure of the rate of information arrival) increases spreads, whereas predictable turnover decreases them (see also Hartmann, 1998a).

29

that, although very appealing, present high computational costs and require
estimation of a large number of parameters (Hull and White, 1987; Scott,
1987; Wiggins, 1987; Chesney and Scott, 1989). An alternative approach,
followed by Jorion (1996), derives ISDs using Fischer Blacks (1976) optionpricing model. Jorion (1996) provides strong evidence for the superiority
of ISDs over time-series models and shows a positive correlation between
volume and volatility that is consistent with the mixture-distribution hypothesis.8
The recent empirical contribution by Ito, Lyons, and Melvin (1998) is,
however, crucial to this literature. Ito, Lyons, and Melvin provide evidence
in favor of the private-information hypothesis, using data for the Tokyo
foreign-exchange market, which from 1972 until 1994 was restricted from
trading over the lunch break (12:00 to 13:30). Following the related study
by French and Roll (1986) for the New York Stock Exchange, Ito, Lyons,
and Melvin start by noting that the three candidate explanations of the
importance of trading for price determination are that (1) public information arrives mainly during trading hours, (2) private information induces
trades that aect prices during trading hours, and (3) errors in pricing are
more likely to occur during trading hours. The authors discriminate among
these explanations by using the following strategy. First, comparing volatility across regimes with an unchanged ow of public information, they nd
that lunch return variance doubles when trading opens. Given that the
foreign-exchange market is largely skewed toward public information, the
fact that public information explains very little is an important nding in
itself.9
Having eliminated public information as the cause of higher volatility,
Ito, Lyons, and Melvin (1998) discriminate between private information
and pricing errors by showing that the volatility U-shape attens over the
full day, because lunch-hour trading induces greater revelation during that
period, leaving smaller shares of information for the morning and afternoon.
The authors also show that the U-shape tilts upward for the day, a trend
suggesting that the private value of the information is temporary, that is,
an open lunch hour reduces the incentive to trade early, because it reduces
the likelihood that prices will reect information before a position can be
opened.10 Further, there appears to be a clear U-shape in the morning
8 Note that the use of ISDs on stock-market data produces less interesting results. The
predictive power of ISDs is, in general, very low and is never higher than the predictive
power of time-series models (Canina and Figlewski, 1993; Lamourex and Lastrapes,
1993).
9 Ito, Lyons, and Melvin (1998) also provide a number of tests to ensure that the ow
of public information was unchanged.
1 0 This is clear, given the denition of private information used by Ito, Lyons, and
Melvin (1998), which includes information that is not common knowledge and that is

30

when Tokyo closes over lunch, but this U-shape disappears after the lunch
opening, as predicted by private-information models. Finally, Ito, Lyons,
and Melvin also nd that the contribution of mispricing to price volatility
is reduced after the lunch hour, a conclusion that coincides with the view
that the increase in lunch variance is caused wholly by mispricing. Overall,
Ito, Lyons, and Melvin (1998) provide the strongest empirical evidence in
the literature in favor of the private-information hypothesis, even though
their studies conclude that private information is expected to predict prices
mainly over relatively short horizons and, thus, may not be fundamental
(see also DeGennaro and Shrieves, 1995, Andersen and Bollerslev, 1998,
and Melvin and Yin, 1999).
Covrig and Melvin (1999) have also tested some implications of marketmicrostructure theory along the lines of Ito, Lyons, and Melvin (1998).
In particular, Covrig and Melvin identify a period in the foreign-exchange
market when there is a high concentration of informed yen-dollar traders
active in Tokyo. Comparing the period of informed-trader clustering to
a similar period without informed traders, they show that exchange-rate
quotes adjust to full-information levels much more quickly when informed
traders are active in the market than when they are not. Covrig and Melvin
also nd that Japanese quotes lead the rest of the market when the informed
traders are active but show a two-way causality when the informed traders
are not active. In addition, the contribution of yen-dollar price discovery
relative to quotes of the rest of the world is 5 to 12 percentage points higher
when the informed traders are active than when they are not. Covrig
and Melvin suggest that their results are consistent with the view that
private information is at times quite important, but that normal times
seem to be those periods during which public information implies a high
contemporaneous correlation across quotes, regardless of the informations
origin. The important implication of this conclusion is that the results
reported earlier by Ito, Lyons, and Melvin are attributable, not to inventory
rebalancing prior to the close, but to private information. This line of
analysis is an important avenue for future research.
Friedmans (1953) classic argument in favor of oating exchange rates
is that rational speculators will, in addition to imparting valuable information to the market, smooth exchange-rate movements, that is, reduce
exchange-rate volatility. In a recent theoretical microstructural analysis
of the connection between rational speculative activity and exchange-rate
volatility, Carlson and Osler (2000) argue that Friedmans analysis implicitly and crucially excludes interest-rate dierentials from his interpretation
price relevant. This denition is less stringent than that used in previous studies (for
example, French and Roll, 1986), which typically require that the price changes induced
by information be permanent.

31

of speculator behavior. They develop a microstructural model that reveals

that informed, rational speculators who consider interest dierentials will
magnify the exchange-rate eects of interest-rate shocks and may thereby
increase overall exchange-rate volatility. This connection between rational speculation and volatility, which does not rely on asymmetric information, is structural, because speculators aect the exchange-rate-generating
process. Carlson and Osler demonstrate that, in their framework, rational
speculation will tend to be stabilizing at low levels of speculative activity
and destabilizing at high levels of activity.

32

6 BID-ASK SPREAD DETERMINATION

The Main Determinants of the Spread
Modeling the bid-ask spread is dicult, because many institutional details
must be considered. Finance theory identies three main determinants of
the bid-ask spread: the cost of dealer services, the cost of adverse selection,
and the cost of holding inventory.
The cost of dealer services has been analyzed formally by Demsetz
(1968), who assumes the existence of some xed costs of predictable immediacy as the service for which compensation is required by market makers.
These costs may include the cost for acquiring expertise in, and subscriptions to, specialized electronic information and trading systems (for example, Reuters).1
The eects of adverse selection on bid-ask spreads were rst examined
by Bagehot (1971), whose model includes both liquidity-motivated transactors, who are willing to pay the price of the spread to the market maker
in exchange for predictable immediacy, and insider transactors, who can
speculate at the expense of the market maker by using insider information.
An adverse-selection problem arises because market makers cannot distinguish between the liquidity-motivated transactors and the insiders and are
thus induced to widen spreads for both categories. The bid-ask spread is
thus the weapon market makers use to defend themselves against adverse
selection. Information-cost models based on the adverse-selection argument
have been built by, among others, Copeland and Galai (1983), Glosten and
Milgrom (1985), and Kyle (1985). Although these models are very appealing, they suer from the same problem as MDH modelsthat is, their
empirical implementation requires the estimation of the rate of information
arrival, or the share of information trading relative to overall trading, and
both of these variables are unobservable.2
1 The traditional view that the centralization of the foreign-exchange market implies
economies of scale such that market makers have a natural monopoly is not supported
by the evidence (Stigler, 1964), even though specialists in the NYSE receive monopoly
rents from other investors in compensation for barriers to market entry (Smidt, 1971).
Barriers to entry or exit in the foreign-exchange market are not generally so great as to
justify the existence of a natural monopoly (Stoll, 1978; Black, 1991).
2 It also follows, from the adverse-selection model, that predictable volume is inversely related to spreads. In fact, given that market makers gain from deals with
liquidity-motivated transactors, spreads decrease as expected order ows from this class
of transactors increase and increase as expected ows decrease (Easley and OHara,
1992).

33

Finally, inventory holding costs are also a determinant of bid-ask spreads.

Holding inventory of foreign exchange may be risky, because the holder is
exposed to market movements in the value of the inventory. Thus, the difference between the bid and ask prices and a theoretical price somewhere
in between may be regarded as a premium required by the inventory holder
as compensation for risk. Inventory-cost models generally assume that market makers optimize their inventory holding. The desired level of inventory
is zero, and a constant spread is shifted continuously, according to the
probability of receiving a purchase and a sale order. In general, inventorycost models imply that market makers shift the spread downward (upward)
and increase the width of the spread when a positive (negative) inventory
is accumulated. The original argument that inventory holding costs are
a crucial determinant of bid-ask spreads derives from Barnea and Logue
(1975), and dynamic-optimization inventory-cost models are presented by
Bradeld (1979), Amihud and Mendelson (1980), and Ho and Stoll (1981).
Modeling the Spread: Theory and Evidence
A peculiarity of the foreign-exchange market is that market making and
brokering are separate functions: market makers do not operate as brokers,
and brokers do not act as market makers. Brokers spreads should thus be
modeled dierently than market makers spreads. A brokered spread is usually the combination of the best bid and ask prices received by the broker as
separate limit orders. Flood (1991) suggests modeling brokered spreads as
a pair of extreme order statistics from independent distributions of buy and
sell limit orders. By denition, the k th order statistic is the k th number in
a list containing the sample realizations of a nite number of independent
random variables ranked in increasing order. Modeling limit orders also
requires the derivation of the distribution of these statistics, which is conditional on volume and on the constraint that the best ask always exceeds
the best bid. Given the analytical complexity of such derivation, Cohen
et al. (1979, 1981) assume a yawl distribution in modeling limit orders
(see also Cohen et al., 1986). Very little work has been done, however, on
modeling brokered spreads, and the underlying theory remains largely to
be developed.
The directly proportional relation between spreads and exchange-rate
volatility is generally accepted by the microstructure literature. Early studies modeled the spread as a function of transactions costs, banks profits from providing liquidity service, and market makers payos for facing
exchange-rate risk by assuming open positions. The main conclusions of
these early studies are that exchange-rate spreads are wider under oating exchange-rate regimes than under xed regimes (Aliber, 1975) and

34

that measures of exchange-rate dispersion (measuring exchange-rate volatility) are followed closely by exchange-rate spreads (Fieleke, 1975; Overturf,
1982).
Glassman (1987) makes a signicant contribution to the literature with
a model that includes variables representing transactions frequency and
considers the nonnormal distribution of exchange-rate change.3 Her model
not only provides additional evidence for the proportional relation between
exchange-rate volatility and bid-ask spreads, it also suggests that market
makers consider a level higher than the second moment of the exchange
rate in order to evaluate the probability of large exchange-rate changes.
Moreover, exchange-rate volatility is predicted by market makers on the
basis of the information provided by both long-run trends (some sixty-ve
days), and very recent experience (one day or one week). Glassman also
nds that spreads widen just before weekends and holidays. Finally, she
shows that transactions costs vary signicantly over time in response to
regime changes on capital controls.
Admati and Peiderer (1988) provide a model that includes three kinds
of agents: informed traders, who have relatively superior information and
only trade on terms favorable to themselves; discretionary liquidity traders,
who must trade sometime during the day but can choose when during the
day to trade in order to minimize costs; and nondiscretionary liquidity
traders, who must trade at a precise time during the day regardless of the
cost. Trading volume is explained by a concentration of trade by informed
traders and discretionary liquidity traders at specic points in time. The
concentrations occur because it is protable for informed traders to trade
when there are many liquidity traders who do not have the same information
they have; discretionary liquidity traders are attracted, because the larger
the number of traders, the lower the cost of trading. The Admati-Peiderer
model also predicts the increase in both volume and volatility that is typical
at the open and close of a trading day, so that trading activity displays a
U-shaped pattern from open to close.
The assumption of traders risk neutrality is, however, crucial in generating the key results of the Admati-Peiderer model. This is rigorously
shown by Subrahmanyam (1991), who models a noncompetitive speculative market in which informed traders, as well as market makers, are risk
averse. His main nding is that market liquidity is nonmonotonic in the
number of informed traders, their degree of risk aversion, and the accuracy
3 The evidence for a nonnormal distribution of exchange-rate change has grown since
the late 1970s and early 1980s (Westereld, 1977; McFarland, Pettit, and Sung, 1982;
Boothe and Glassman, 1987). The distribution appears to be quite close to a leptokurtic
pattern, with higher peaks and fatter tails than the normal distribution. Because nonnormal distributions are not described completely by the rst two moments, Glassman
(1987) includes the rst through fourth moments in her model.

35

of their information. His model also predicts that price eciency is reduced
by increased concentration of liquidity traders, and that market liquidity
may also be nonmonotonic in the variance of liquidity traders.
Bollerslev and Domowitz (1993) use intradaily data to investigate the
behavior of quote arrivals and bid-ask spreads over the trading day, across
geographical locations, and across market participants. Their ndings are
useful for discriminating among theoretical models of trading activity. In
particular, they determine that trading activity and the bid-ask spreads for
traders whose activity is restricted to regional markets can be described
by a U-shaped distribution consistent with the Admati-Peiderer (1988)
predictions. The patterns of trading activity and spreads during the day,
however, also strongly suggest a degree of risk aversion by traders, a nding
that is consistent with Subrahmanyams model: given a measure of risk
aversion, the more trading that is executed by informed participants, the
higher will be its cost.
Goodhart and Figliuoli (1991) study minute-by-minute spot rates (bidask Reuters quotes) on three days in 1987 (September 14, and 15, October
21) at the Reuters screen in London. They nd evidence that leptokurtosis
and heteroskedasticity are less pronounced at minute-by-minute frequencies
than at lower frequencies. In addition, leptokurtosis, skewness, and heteroskedasticity are time-varying. Trading volume is also time-varying and
is higher at the European and North American market openings and lower
at the European lunch hour. The series also exhibit a rst-order negative
serial correlation that is especially pronounced immediately after jumps in
the exchange rate. Time aggregation appears to reduce, but not eliminate,
the rst-order autocorrelation. Finally, multivariate analysis suggests that
there is a signicant link between lagged exchange rates (both the domestic
rate and the deutsche mark-dollar exchange rate) and the current spot rate
(see also Goodhart, Ito, and Payne, 1996).
Bollerslev and Melvin (1994), using an asymmetric-information model
with informed traders and liquidity traders in the tradition of Glosten
and Milgrom (1985) and Admati and Peiderer (1988), show that bidask spreads are proportionally related to exchange-rate uncertainty. Their
studys innovation is that they employ an ordered probit analysis in order
to capture the discreteness in the spread distribution, with the uncertainty
of the spot exchange rate modeled as a GARCH process (see also Melvin
and Ramirez, 2000).
Overall, the main ndings of the literature on the behavior of bid-ask
spreads in the foreign-exchange market during the 1980s and early 1990s
may be summarized by stating that spreads are directly proportional to
the volatility of exchange rates and trading volume, and that they are
higher on Fridays. In a more recent study, Bessembinder (1994) nds that

36

bid-ask spreads are also proportionally related to forecasts of inventory

price risk. Bessembinder conrms that spreads widen before weekends and
other nontrading intervals but explains that this pattern reects an increase
in the sensitivity of spreads to risk and liquidity costs over nontrading
intervals. He uses a regression of time series of currency spreads on proxies
for inventory cost and trading volume to study the behavior of bid-ask
spreads and uses the generalized method of moments to make estimations.4
Moreover, because the bid and ask prices quoted by market makers
are not necessarily symmetric around the underlying value of the asset,
Bessembinder develops a simple procedure for estimating the location of
bid and ask quotes with respect to value. The main innovation in his
method is that it does not require the simultaneous observation of spot and
forward prices that previous attempts in the literature employ (for example,
Bossaerts and Hillion, 1991). Bessembinders method is derived as follows.
He denes an unobservable value Wt as
Wt = t At + (1 t ) Bt ,

(16)

where t is a time-t location parameter, At denotes the ask quote, and Bt

denotes the bid quote. He then denes the change in the underlying value
of the currency as
(17)
Wt+1 = t + t+1 ,
where t denotes the expectation at time t of the change in value over the
next period and t+1 is the unexpected change. Combining (16) and (17)
yields:
Bt+1 = t + t+1 (St+1 ) + t St + t+1 ,
(18)
where St At Bt denotes the bid-ask spread. In addition, the two
time-varying parameters t and t are assumed to be linear functions of
observable variables Xt and Zt , respectively. Thus, equation (18) becomes
Bt+1 = 0 + 1 Xt + 0 [St+1 ] + 1 [ (Zt+1 St+1 )] + t+1 .

(19)

In estimating equation (19), Bessembinder (1994) includes in Xt a Monday indicator variable and the excess of thirty-day eurodollar interest rates
over thirty-day local-currency euro rates; he considers in Zt a Friday indicator, a preholiday indicator, the change since the preceding day in the
thirty-day eurodollar deposit rate, and the change since the preceding day
in the local-currency euro interest rate. The results strongly suggest that
currency market makers reduce quotes in relation to the underlying dol4 Bessembinder uses forecasts of price risk, interest-rate-based measures of liquidity
costs, and a nontrading indicator as proxies for inventory carrying costs. He uses the
forecastable and unexpected components of futures trading volume as proxies for trading
volume.

37

lar value when U.S. interest rates are rising and also, less signicantly, on
Fridays. The results also suggest a Monday eect on currency values, but
shifts in the placement of quotes relative to value as weekends approach
make it dicult to detect. In addition, Bessembinders (1994, p. 344) evidence illustrates that inference regarding asset value can be altered by
allowing for variation in the placement of quotes in relation to value.
Lee (1994) provides an alternative modeling strategy for the conditional
heteroskedasticity of the prediction error of foreign-exchange rates. Under
the assumption of cointegration between spot and forward rates, Lee uses
a system of error-correction models for GARCH-type models as a function of the spread. Estimating the system using daily series for seven exchange rates, he nds a strong correlation between spreads and exchangerate volatility.5
Jorion (1996) uses an option-implied volatility in his model specication
of bid-ask spreads. His results are generally consistent with the implications
of conventional spread theory. He (1996) also shows that ISDs dominate
all other risk measures in explaining bid-ask spreads.
Bessembinder, Chan, and Seguin (1996) investigate the relation between
trading volume and some proxies for information ows (ratio of volatility of
returns to a diversied equity portfolio) and for divergencies in opinion (the
open interest of the Standard & Poors 500 Index futures contract). They
nd that, in both spot and futures markets, trading volume varies positively
with the proxies for information ows. The choice of the trading venue
largely depends, as one would expect, on the nature of information ows:
traders informed about rm-specic matters trade primarily in the spot equity market, whereas traders with more general, market-wide information
choose to trade in the spot market. Bessembinder, Chan, and Seguin (1996)
also nd that trading volumes in both spot and futures markets are positively related to the proxy for divergencies of opinion when these volumes
rise but are unrelated to the proxies when they fall. The day-of-the-week
eects appear to be asymmetric across markets, with lower futures volume
relative to spot volume occurring late in the week (Foster and Viswanathan,
1990, develop a theoretical model that predicts this asymmetry). Bessembinder, Chan, and Seguin (1996, p. 132) conclude that additional research
is warranted on identifying those circumstances under which price forma5 Another paper relating global foreign-exchange markets to spreads is that by Hartmann (1998b), who estimates the long-run impact of trading activity on bid-ask spreads
by using a short panel containing around-the-clock Reuters quotes and global transactions volumes; allowance is made for individual and time eects in an unbalanced
random-eects model. Hartmann nds that the volume parameter has a (weakly) statistically signicant negative sign, in line with liquidity-eect explanations, and that the
volatility parameter is positive. He nds that structural parameters are stable over time
and that residuals are, as a group, heteroskedastic.

38

tion will depend primarily on order ow or on the observation of public

information.6
Hsieh and Kleidon (1996) stress the empirical diculties in reconciling
the implications of asymmetric-information models with the observed timeseries behavior of stock- and foreign-exchange data, unless liquidity traders
demand to trade is very high at market open and close for reasons that are
not explained by standard asymmetric-information models. One important
inconsistency lies in the fact that bid-ask spreads are observed to go up,
rather than down, at both open and close, in contradiction to the implication of asymmetric-information models that liquidity traders are trading
less at these times because transactions costs are higher. Another empirical
regularity that is not explained by the traditional spread theory is that neither trading volume nor volatility in the New York foreign-exchange market
is signicantly aected at market close in London, when both volume and
volatility are at a peak in that market. Hsieh and Kleidon give two dierent explanations for this empirical failure of information-based models of
bid-ask spread determination. First, traders are known to learn the feel
of the market, that is, they go through a learning process. This learning
process, until now largely ignored in spread theory, may help explain the
high trading volume, volatility, and spreads in the morning. In addition,
the peaks of trading volume, volatility, and spreads at the close of foreignexchange markets may be explained by the desire of traders to unload excess
inventories. This remains an interesting area for further investigation.

6 Brock and Kleidon (1992) examine the eect of periodic stock-market closures on
transactions demand and volume of trade and, consequently, bid and ask prices. They
nd that transactions demand is greater and less elastic at market open and close than at
other times during the trading day. A market maker such as a New York Stock Exchange
specialist may thus eectively discriminate by charging a higher transaction price during
these periods of peak demand. The predictions of periodic demand with high volume and
concurrent wide spreads are consistent with empirical evidence, whereas the predictions
of current information-based models are not.

39

7 MARKET MAKERS BEHAVIOR

The microstructure literature has also examined the way in which market makers behavior aects the eciency of the market. The two most
common concerns with respect to their behavior are the determination of
bid-ask spreads and the treatment of price information. Chapter 6 discussed
the former issue; this chapter discusses the latter. An extensive literature
already exists for equity markets, but the foreign-exchange market is receiving increasing interest, led, notably, by Lyons (1991, 1995, 1996, 1998,
2001).
Private Beliefs and Information Externalities
Lyons (1991) examines the importance of private beliefs and information
externalities in the foreign-exchange market. He argues that the absence of
dierent beliefs about the exchange-rate path, as well as the process for updating these beliefs in standard exchange-rate-determination models, may
be crucial factors in explaining the empirical failure of such models. He
(Lyons, 1991, pp. 1-2) examines the way in which the transactions themselves inuence beliefs over time, the eect that the markets institutional
features have on patterns of trading volume, and the mechanisms that the
market has for information transmission:
Agents transact because they dier. It is important to distinguish
between dierences in valuation beliefs and dierences that arise for
other reasons. Of course, it is possible for investors to agree on
valuation and still choose to transact for diversication purposes.
However, in the context of the typical portfolio choice models, the
trading volume that can be explained as a result of actual shifts in
wealth, taxes, return second moments, etc., is minute in comparison
to actual volumes, and in the simplest models is zero. The burden
of explanation, then, appears almost certainly to fall on dierences
in beliefs regarding valuation.

Most models in the literature describe equilibrium-price determination

as a one-shot outcome from the activity of traders who use private information as well as the information in market-clearing prices to determine their
demands for risky assets (Diamond and Verrecchia, 1981). Lyons (1991),
by allowing the transactions themselves to aect the updating process of
private beliefs directly over time, suggests that the equilibrium is no longer
a one-shot outcome, but the outcome, instead, of an interaction among

40

private beliefs, volume, and volatility. This interactive process leads, however, to a sort of information externality caused by the dual role of traders
as speculators, on the one hand, and as information clearinghouse (intermediating customer orders that contain information), on the other hand.1
Prot maximization induces traders to underestimate important information in making their trading decisions, thereby reducing the information
content of prices at any given time. A crucial implication of this model is
that the greater the market power and the degree of traders risk aversion,
the less the information that will be revealed by market prices.
Lyons (1995) executes a number of tests on standard microstructure
hypothesis in the foreign-exchange market. In order to make these tests, he
builds a model based on realistic assumptions of the institutional features of
the foreign-exchange market. Lyons records the transactions of one dealer
and one broker in the U.S. market for ve days during August 1992. The
data set includes three interlocking components: the direct quotes and
trades of a market maker from a leading New York bank, the position
cards of the same market maker, and the prices and quantities for thirdparty transactions intermediated by a leading New York dealer. The main
focus of the study is to test for eects of trading volume on quoted prices
through the two channels emphasized by the literature: the information
channel and the inventory-control channel. The results suggest that trading
volume aects quoted prices through both channels, therefore giving some
support to both strands of the microstructure literature.2
Event Uncertainty versus Hot Potato
Lyons (1996) extends his 1995 model and uses the same data set to shed
light on the statistical relation between the intensity of trading and the informativeness of trades. In particular, he attempts to discriminate between
the two polar views of trading intensity that he describes as (1) the eventuncertainty view (trades are more informative when trading intensity is
high) and (2) the hot potato view (trades are more informative when
trading intensity is low). The event-uncertainty hypothesis can be traced
back to Easley and OHara (1992), who build a model in which new information may not exist (in contrast to most of the asymmetric-information
models in this literature). In such a framework, there is a probability p
of new information and (1 p) of no new information; also, the new information is good news with probability q and bad news with probability
1 Information externalities are predicted by earlier theoretical models under dierent
mechanisms (Stein, 1987).
2 In modeling the price-expectation formation process, Lyons uses a Bayesian model
in the tradition of Amihud and Mendelson (1980), Cohen et al. (1981), Conroy and
Winkler (1981), Glosten and Milgrom (1985), and Madhavan and Smidt (1991). For a
nontechnical exposition of these models, see Flood (1991).

41

(1 q). Easley and OHara (1992) show that trades occurring when intensity is high should induce a larger updating of beliefs than when trading
intensity is low, implying that trades are more informative when trading
intensity is high. The opposite view, the hot potato view, is rationalized on
the basis of asymmetric-information models in the tradition of Admati and
Peiderer (1988), in which liquidity traders clump together in their trading
in order to minimize their losses to informed traders. Because of this clumping of liquidity traders, trades are more informative when trading intensity
is low.3 The main goal of Lyons (1996) is to examine whether currencytrading volume is informative and under what circumstances. Specically,
he uses transactions data to test whether trades occurring when trading intensity is high are more informative, dollar for dollar, than trades occurring
when intensity is low. Lyons empirical results in estimating his model are
supportive of the one or the other hypothesis depending upon the measure
of trading intensity used in estimation, implying that taken together, the
results highlight the potential complementarity between these seemingly
polar views (Lyons, 1996, p. 20).
Lyons (1998) also examines foreign-exchange trading at the dealer level.
The dealer Lyons tracks averages U.S. $100,000 in prots per day on a volume of U.S. $1 billion per day. The half-life of the dealers position is about
ten minutes, which may be interpreted as evidence supporting inventory
models. The author also identies the dealers speculative position over
time and nds that this position determines the share of prots deriving
from speculation, rather than intermediation, and that intermediation has
a relatively more important role.

3 See

also Lyons (1997), who develops a simultaneous-trade model of the spot foreignexchange market that produces hot-potato trading. At the outset, risk-averse dealers
receive customer orders that are not generally observable. Dealers then trade among
themselves. Thus, each dealer intermediates both his or her customers trades and any
information contained therein. This information is subsequently revealed in price, depending on the information in interdealer trades. Lyons shows that hot-potato trading
reduces the information in interdealer trades, making price less informative.

42

8 CONCLUSION
We began this study by noting that the failure of fundamentals-based
exchange-rate models to explain or predict exchange-rate movements reliably has, at least in part, motivated the development of the microstructure literature (Flood and Taylor, 1996). To date, however, the foreignexchange-market microstructure seems to shed light most strongly on related issues, such as the transmission of information among market participants, the heterogeneity of agents expectations, and the implications of
agents heterogeneity for trading volume and exchange-rate volatility.1
There appears to be an emerging consensus in the literature, moreover,
that macroeconomic fundamentals are a reasonable guide to very long-run
exchange-rate movements (Flood and Taylor, 1996; Lothian and Taylor,
1996), and there is increasing interest in exchange-rate models with rigorous, stochastic, general-equilibrium microeconomic foundations (Obstfeld
and Rogo, 1995, 1996; Lane, 1999; Sarno, 2001). For the foreseeable future, therefore, it seems that research in foreign-exchange markets is likely
to be dominated by three strands: long-run empirical exchange-rate modeling, the new open-economy macroeconomics, and analyses of the foreignexchange-market microstructure. Synthesizing these three approaches into
a unied treatment would seem to be a worthwhile challenge for the economics profession.

1 An important exception is the recent paper by Evans and Lyons (1999), which
demonstrates that order ow may empirically explain a large portion of nominal
exchange-rate movements over periods of four months or less; this nding clearly suggests the need for further research on the relation between order ow and exchange-rate
movements.

43

REFERENCES
Admati, Anat R., and Paul Pfleiderer, A Theory of Intraday Patterns: Volume
and Price Variability, Review of Financial Studies, 1 (No. 1, 1988), pp. 340.
Aliber, Robert Z., Monetary Independence under Floating Exchange Rates,
Journal of Finance, 30 (No. 2, 1975), pp. 365376.
Allen, Helen, and Mark P. Taylor, Charts, Noise and Fundamentals in the
London Foreign Exchange Market, Economic Journal, 100, Supplement (No.
400, 1990), pp. 4959.
Amihud, Yakov, and Haim Mendelson, Dealership Market: MarketMaking
with Inventory, Journal of Financial Economics, 8 (No. 1, 1980), pp. 3153.
Andersen, Torben G., and Tim Bollerslev, DM-Volatility: Intraday Activity
Patterns, Macroeconomic Announcements, and Longer Run Dependencies,
Journal of Finance, 53 (No. 1, 1998), pp. 219265.
Bagehot, Walter, The Only Game in Town, Financial Analysts Journal, 22
(No. 1, 1971), pp. 1214.
Baillie, Richard T., Econometric Tests of Rationality and Market Efficiency,
Econometric Reviews, 8 (No. 2, 1989), pp. 151186.
Baillie, Richard T., and Tim Bollerslev, The Message in Daily Exchange Rates:
A Conditional-Variance Tale, Journal of Business and Economic Statistics,
7 (No. 3, 1989), pp. 297305.
, A Multivariate Generalized ARCH Approach to Modeling Risk Premia
in Forward Foreign Exchange Rate Markets, Journal of International Money
and Finance, 9 (No. 3, 1990), pp. 309324.
, Intraday and Inter-Market Volatility in Foreign Exchange Rates,
Review of Economic Studies, 58 (No. 3, 1991), pp. 565585.
Bank for International Settlements (BIS), Central Bank Survey of Foreign
Exchange and Derivatives Market Activity, Basle, Bank for International
Settlements, 1998.
Barnea, Amir, and Dennis E. Logue, The Effect of Risk on the Market
Makers Spread, Financial Analysts Journal, 26 (No. 1, 1975), pp. 4549.
Bessembinder, Hendrik, Bid-Ask Spreads in the Interbank Foreign Exchange
Markets, Journal of Financial Economics, 35 (No. 3, 1994), pp. 317348.
Bessembinder, Hendrik, Kalok Chan, and Paul J. Seguin, An Empirical
Examination of Information, Differences of Opinion, and Trading Activity,
Journal of Financial Economics, 40 (No. 1, 1996), pp. 105134.
Bilson, John F. O., Technical Currency Trading, in Lee R. Thomas, ed., The
Currency-Hedging Debate, London, IFR Publishing, 1990, pp. 257275.
Black, Fischer, The Pricing of Commodity Contracts, Journal of Financial
Economics, 3 (No. 1/2, 1976), pp. 167179.

44

Black, Stanley W., Transactions Costs and Vehicle Currencies, Journal of

International Money and Finance, 10 (No. 4, 1991), pp. 512526.
Bollerslev, Tim, Generalized Autoregressive Conditional Heteroskedasticity,
Journal of Econometrics, 31 (No. 3, 1986), pp. 307327.
, Modelling the Coherence in Short-Run Nominal Exchange Rates: A
Multivariate Generalized ARCH Model, Review of Economics and Statistics,
72 (No. 3, 1990), pp. 498505.
Bollerslev, Tim, Ray Y. Chou, and Kenneth F. Kroner, ARCH Modeling in
Finance: A Review of the Theory and Empirical Evidence, Journal of
Econometrics, 52 (No. 12, 1992), pp. 559.
Bollerslev, Tim, and Ian Domowitz, Trading Patterns and Prices in the
Interbank Foreign Exchange Market, Journal of Finance, 48 (No. 4, 1993),
pp. 14211443.
Bollerslev, Tim, and Michael T. Melvin, Bid-Ask Spreads and Volatility in the
Foreign Exchange Market: An Empirical Analysis, Journal of International
Economics, 36 (No. 34, 1994), pp. 355372.
Boothe, Paul M., and Debra A. Glassman, The Statistical Distribution of
Exchange Rates: Empirical Evidence and Economic Implications, Journal
of International Economics, 22 (No. 34, 1987), pp. 297319.
Bossaerts, Peter, and Pierre Hillion, Market Microstructure Effects of
Government Intervention in the Foreign Exchange Market, Review of
Financial Studies, 4 (No. 3, 1991), pp. 513541.
Bradfield, James, A Formal Dynamic Model of Market Making, Journal of
Financial and Quantitative Analysis, 14 (No. 2, 1979), pp. 275291.
Brock, William A., and Allan W. Kleidon, Periodic Market Closure and
Trading Volume: A Model of Intraday Bids and Asks, Journal of Economic
Dynamics and Control, 16 (No. 34, 1992), pp. 451489.
Campbell, John Y., Andrew W. Lo, and A. Craig MacKinlay, The Econometrics
of Financial Markets, Princeton, N.J., Princeton University Press, 1997.
Canina, Linda, and Stephen Figlewski, The Informational Content of Implied
Volatility, Review of Financial Studies, 6 (No. 3, 1993), pp. 659681.
Carlson, John A., and Carol L. Osler, Rational Speculators and Exchange Rate
Volatility, European Economic Review, 44 (No. 2, 2000), pp. 231253.
Chesney, Marc, and Louis Scott, Pricing European Currency Options: A
Comparison of the Modified Black-Scholes Model and a Random Variance
Model, Journal of Financial and Quantitative Analysis, 24 (No. 3, 1989),
pp. 267284.
Cheung, Yin-Wong, and Menzie D. Chinn, Macroeconomic Implications of the
Beliefs and Behavior of Foreign Exchange Traders, Department of Economics, University of California at Santa Cruz, 1999, processed.
Cheung, Yin-Wong, Menzie D. Chinn, and Ian W. Marsh, How Do UK-Based Foreign Exchange Dealers Think Their Market Operates? CEPR
Discussion Paper No. 2230, London, Centre for Economic Policy Research,
September 1999.

45

Cheung, Yin-Wong, and Clement Yuk-Pang Wong, Foreign Exchange Traders

in Hong Kong, Tokyo and Singapore: A Survey Study, Advances in Pacific
Basin Financial Markets, 5 (No. 1, 1999), pp. 111134.
, A Survey of Market Practitioners: Views on Exchange Rate Dynamics,
Journal of International Economics, 51 (No. 2, 2000), pp. 401419.
Clark, Peter K., A Subordinated Stochastic Process Model with Finite Variance
for Speculative Prices, Econometrica, 41 (No. 1, 1973), pp. 135156.
Cohen, Kalman J., Steven F. Maier, Robert A. Schwartz, and David K. Whitcomb, On The Existence of Serial Correlation in an Efficient Securities
Market, TIMS Studies in the Management Sciences, 11 (No. 2, 1979), pp.
151168.
, Transactions Costs, Order Placement Strategy and Existence of the
Bid-Ask Spread, Journal of Political Economy, 89 (No. 2, 1981), pp.
287305.
, The Microstructure of Securities Markets, Englewood Cliffs, N.J.,
Prentice-Hall, 1986.
Conroy, Robert M., and Robert L. Winkler, Informational Differences between
Limit and Market Orders for a Market Maker, Journal of Financial and
Quantitative Analysis, 16 (No. 5, 1981), pp. 703724.
Copeland, Thomas E., and Dan Galai, Information Effects on the Bid-Ask
Spread, Journal of Finance, 38 (No. 5, 1983), pp. 14571469.
Cornell, Bradford, The Consumption Based Asset Pricing Model: A Note on
Potential Tests and Applications, Journal of Financial Economics, 9 (No. 1,
1981), pp. 103108.
Covrig, Vicentiu, and Michael T. Melvin, Asymmetric Information and Price
Discovery in the FX Market: Does Tokyo Know More About the Yen?
Department of Economics, Arizona State University, Tempe, Ariz., 1999,
processed.
Curcio, Riccardo, and Charles A. E. Goodhart, When Support/Resistence
Levels Are Broken, Can Profits Be Made? Evidence from the Foreign
Exchange Market? London School of Economics, Financial Markets Group
Discussion Paper No. 142, July 1992, pp. 122.
, Chartism: A Controlled Experiment, Journal of International Securities
Markets, 7 (No. 2, 1993), pp. 173186.
DeGennaro, Ramon P., and Ronald E. Shrieves, Public Information Releases,
Private Information Arrival, and Volatility in the Foreign Exchange Markets,
paper presented at Olsen and Associates Conference on High-Frequency
Data in Finance, in Zurich, March 2527, 1995.
De Grauwe, Paul, and Hans Dewachter, A Chaotic Model of the Exchange
Rate: The Role of Fundamentalists and Chartists, Open Economies Review,
4 (No. 4, 1993), pp. 351379.
De Grauwe, Paul, Hans Dewachter, and Mark Embrechts, Exchange Rate
Theory: Chaotic Models of Foreign Exchange Markets, Oxford and Cambridge, Mass., Blackwell, 1993.

46

De Long, J. Bradford, Andrei Shleifer, Lawrence H. Summers, and Robert J.

Waldmann, The Size and Incidence of the Losses from Noise Trading,
Journal of Finance, 44 (No. 3, 1989), pp. 681699.
, Noise Trader Risk in Financial Markets, Journal of Political Economy,
98 (No. 4, 1990a), pp. 703738.
, Positive Feedback Investment Strategies and Destabilizing Rational
Speculation, Journal of Finance, 45 (No. 2, 1990b), pp. 379395.
Demos, Antonis A., and Charles A. E. Goodhart, The Interaction between the
Frequency of Market Quotations, Spread and Volatility in the Foreign
Exchange Markets, Applied Economics, 28 (No. 3, 1996), pp. 377386.
Demsetz, Harold, The Cost of Transacting, Quarterly Journal of Economics,
82 (No. 1, 1968), pp. 3353.
Diamond, Douglas W., and Robert E. Verrecchia, Information Aggregation in
a Noisy Rational Expectations Economy, Journal of Financial Economics, 9
(No. 3, 1981), pp. 221235.
Diebold, Francis X., Serial Correlation and the Combination of Forecasts,
Journal of Business and Economic Statistics, 6 (No. 1, 1988), pp. 105111.
Diebold, Francis X., and James A. Nason, Nonparametric Exchange Rate
Prediction? Journal of International Economics, 28 (No. 34, 1990), pp.
315332.
Diebold, Francis X., and Marc Nerlove, The Dynamics of Exchange Rate
Volatility: A Multivariate Latent Factor ARCH Model, Journal of Applied
Econometrics, 4 (No. 1, 1989), pp. 121.
Diebold, Francis X., and Peter Pauly, Has the EMS Reduced Member-Country
Exchange Rate Volatility? Empirical Economics, 13 (No. 2, 1988), pp.
81102.
Dominguez, Kathryn M.E., Exchange Rate Efficiency and the Behavior of
International Asset Markets, Ph.D. dissertation, Yale University, 1987.
Domowitz, Ian, and Craig S. Hakkio, Conditional Variance and the Risk
Premium in the Foreign Exchange Market, Journal of International
Economics, 19 (No. 12, 1985), pp. 4766.
Dornbusch, Rudiger, Expectations and Exchange Rate Dynamics, Journal of
Political Economy, 84 (No. 6, 1976), pp. 11611176.
Dubofsky, David A., Options and Financial Futures: Valuation and Uses, New
York, McGraw-Hill, 1992.
Easley, David, and Maureen OHara, Time and the Process of Security Price
Adjustment, Journal of Finance, 47 (No. 2, 1992), pp. 577605.
Engle, Robert F., Autoregressive Conditional Heteroscedasticity with Estimates of the Variance of United Kingdom Inflation, Econometrica, 50 (No.
4, 1982), pp. 9871007.
Engle, Robert F., and Tim Bollerslev, Modelling the Persistence of Conditional Variances, Econometric Reviews, 5 (No. 1, 1986), pp. 150.
Engle, Robert F., Takatoshi Ito, and Wen-Ling Lin, Meteor Showers or Heat
Waves? Heteroskedastic Intra-Daily Volatility in the Foreign Exchange
Market, Econometrica, 58 (No. 3, 1990), pp. 525542.

47

Engle, Robert F., David M. Lilien, and Russell P. Robins, Estimating Time
Varying Risk Premia in the Term Structure: The Arch-M Model, Econometrica, 55 (No. 2, 1987), pp. 391407.
Epps, Thomas W., and Mary Lee Epps, The Stochastic Dependence of
Security Price Changes and Transaction Volumes: Implications for the
Mixture-of-Distributions Hypothesis, Econometrica, 44 (No. 2, 1976), pp.
305321.
Evans, Martin D. D., and Richard K. Lyons, Order Flow and Exchange Rate
Dynamics, National Bureau of Economic Research Working Paper No. 7317,
Cambridge, Mass., National Bureau of Economic Research, 1999.
Fama, Eugene, Efficient Capital Markets: A Review of Theory and Empirical
Work, Journal of Finance, 25 (No. 2, 1970), pp. 383417.
Fieleke, Norman S., Exchange-Rate Flexibility and the Efficiency of the
Foreign-Exchange Markets, Journal of Financial and Quantitative Analysis,
10 (No. 3, 1975), pp. 409428.
Flood, Mark D., Microstructure Theory and the Foreign Exchange Market,
Federal Reserve Bank of St. Louis Review, 73 (No. 6, 1991), pp. 5270.
, Market Structure and Inefficiency in the Foreign Exchange Market,
Journal of International Money and Finance, 13 (No., 2, 1994), pp. 131158.
Flood, Robert P., and Mark P. Taylor, Exchange Rate Economics: Whats
Wrong with the Conventional Macro Approach? in Frankel, Galli, and
Giovannini, eds., The Microstructure of Foreign Exchange Markets, 1996, pp.
261301.
Foster, F. Douglas, and S. Viswanathan, A Theory of the Interday Variations
in Volume, Variance, and Trading Costs in Securities Markets, Review of
Financial Studies, 3 (No. 4, 1990), pp. 593624.
Frankel, Jeffrey A., and Kenneth A. Froot, Understanding the U.S. Dollar in
the Eighties: The Expectations of Chartists and Fundamentalists, Economic
Record, 0, Supplement (No. 0, 1986), pp. 2438.
, Using Survey Data to Test Standard Propositions Regarding Exchange
Rate Expectations, American Economic Review, 77 (No. 1, 1987), pp.
133153.
, Chartists, Fundamentalists, and Trading in the Foreign Exchange
Market, American Economic Review, 80 (No. 2, 1990a), pp. 181185.
, Chartists, Fundamentalists and the Demand for Dollars, in Anthony
Courakis and Mark P. Taylor, eds., Private Behaviour and Government Policy
in Interdependent Economies, Oxford, Clarendon, 1990b, pp. 73126.
Frankel, Jeffrey A., Giampaolo Galli, and Alberto Giovannini, eds., The
Microstructure of Foreign Exchange Markets, Chicago, Ill., Chicago University Press for National Bureau of Economic Research, 1996.
Frankel, Jeffrey A., and Andrew K. Rose, Empirical Research on Nominal
Exchange Rates, in Gene M. Grossman, and Kenneth Rogoff, eds., Handbook of International Economics, Vol. 3, Amsterdam and New York, NorthHolland, Elsevier, 1995, pp. 16891729.

48

Friedman, Milton, The Case for Flexible Exchange Rates, in Friedman,

Essays in Positive Economics, Chicago, Ill., University of Chicago Press,
1953, pp. 157203.
French, Kenneth, and Richard Roll, Stock Return Variances: The Arrival of
Information and the Reaction of Traders, Journal of Financial Economics,
17 (No. 1, 1986), pp. 330.
Froot, Kenneth A., and Takatoshi Ito, On the Consistency of Short-Run and
Long-Run Exchange Rate Expectations, Journal of International Money and
Finance, 8 (No. 4, 1989), pp. 487510.
Garbade, Kenneth D., The Effect of Interdealer Brokerage on the Transnational Characteristics of Dealer Markets, Journal of Business, 51 (No. 3,
1978), pp. 477498.
Garbade, Kenneth D., Jay L. Pomrenze, and William L. Silber, On the
Information Content of Prices, American Economic Review, 69 (No. 1,
1979), pp. 5059.
Garman, Mark B., Market Microstructure, Journal of Financial Economics, 3
(No. 3, 1976), pp. 257275.
Giovannini, Alberto, and Philippe Jorion, The Time Variation of Risk and
Return in the Foreign Exchange and Stock Markets, Journal of Finance, 44
(No. 2, 1989), pp. 307325.
Glassman, Debra A., Exchange Rate Risk and Transactions Costs: Evidence
from Bid-Ask Spreads, Journal of International Money and Finance, 6 (No.
4, 1987), pp. 479490.
Glosten, Lawrence R., Is the Electronic Open Limit Order Book Inevitable?
Journal of Finance, 49 (No. 4, 1994), pp. 11271161.
Glosten, Lawrence R., and Paul R. Milgrom, Bid, Ask and Transaction Prices
in a Specialist Market with Heterogeneously Informed Traders, Journal of
Financial Economics, 14 (No. 1, 1985), pp. 71100.
Goldberg, Linda S., and Rafael Tenorio, Strategic Trading in a Two-Sided
Foreign Exchange Auction, Journal of International Economics, 42 (No. 34,
1997), pp. 299326.
Goldberg, Michael D., and Roman Frydman, Theories, Consistent Expectations and Exchange Rate Dynamics, in Helmut Frisch and Andreas Worgotter, eds., Open-Economy Macroeconomics: Proceedings of a Conference Held
in Vienna by the International Economic Association, New York, St. Martins
in association with the International Economic Association, 1993, pp.
377399.
Goodhart, Charles A. E., The Foreign Exchange Market: A Random Walk with
a Dragging Anchor, Economica, 55 (No. 220, 1988), pp. 437460.
Goodhart, Charles A. E., and L. Figliuoli, Every Minute Counts in Financial
Markets, Journal of International Money and Finance, 10 (No. 1, 1991), pp.
2352.
Goodhart, Charles A. E., and Marcello M. Giugale, From Hour to Hour in the
Foreign Exchange Market, Manchester School of Economic and Social
Studies, 61 (No. 1, 1993), pp. 134.

49

Goodhart, Charles A. E., Takatoshi Ito, and Richard Payne, One Day in June
1993: A Study of the Working of the Reuters 20002 Electronic Foreign
Exchange Trading System, in Frankel, Galli, and Giovannini, eds., The
Microstructure of Foreign Exchange Markets, 1996, pp. 107179.
Grammatikos, Theoharry, and Anthony Saunders, Futures Price Variability: A
Test of Maturity and Volume Effects, Journal of Business, 59 (No. 2, 1986),
pp. 319330.
Group of Thirty, The Foreign Exchange Market in the 1980s, Group of Thirty,
New York, 1985.
Hakkio, Craig S., Expectations and the Forward Exchange Rate, International
Economic Review, 22 (No. 3, 1981), pp. 663678.
Harris, Lawrence, A Transaction Data Study of Weekly and Intradaily Patterns
in Stock Returns, Journal of Financial Economics, 16 (No. 1, 1986), pp.
99117.
Hartmann, Philipp, Currency Competition and Foreign Exchange Markets: The
Dollar, the Yen and the Euro, Cambridge, New York, and Melbourne,
Cambridge University Press, 1998a.
, Do Reuters Spreads Reflect Currencies Differences in Global Trading
Activity? Journal of International Money and Finance, 17 (No. 5, 1998b),
pp. 757784.
, Trading Volumes and Transaction Costs in the Foreign Exchange
Market: Evidence from Daily Dollar-Yen Spot Data, Journal of Banking and
Finance, 23 (No. 5, 1999), pp. 801824.
Harvey, Campbell R., and Roger D. Huang, Volatility in the Foreign Currency
Futures Market, Review of Financial Studies, 4 (No. 3, 1991), pp. 543569.
Ho, Thomas, and Hans R. Stoll, Optimal Dealer Pricing under Transactions
and Return Uncertainty, Journal of Financial Economics, 9 (No. 1, 1981),
pp. 4773.
Hodrick, Robert J., Risk, Uncertainty, and Exchange Rates, Journal of
Monetary Economics, 23 (No. 3, 1989), pp. 433459.
Hodrick, Robert J., and Sanjay Srivastava, An Investigation of Risk and Return
in Forward Foreign Exchange, Journal of International Money and Finance,
3 (No. 1, 1984), pp. 529.
Hogan, Kedreth C., Jr., and Michael T. Melvin, Sources of Meteor Showers
and Heat Waves in the Foreign Exchange Market, Journal of International
Economics, 37 (No. 34, 1994), pp. 239247.
Hsieh, David A., The Statistical Properties of Daily Foreign Exchange Rates:
19741983, Journal of International Economics, 24 (No. 12, 1988), pp.
129145.
, Testing for Nonlinear Dependence in Daily Foreign Exchange Rates,
Journal of Business, 62 (No. 3, 1989a), pp. 339368.
, Modeling Heteroscedasticity in Daily Foreign-Exchange Rates, Journal
of Business and Economic Statistics, 7 (No. 3, 1989b), pp. 307317.
Hsieh, David A., and Allan W. Kleidon, Bid-Ask Spreads in Foreign Exchange
Markets: Implications for Models of Asymmetric Information, in Frankel,

50

Galli, and Giovannini, eds., The Microstructure of Foreign Exchange Markets,

1996, pp. 4165.
Hull, John C., and Alan White, Hedging the Risks from Writing Foreign
Currency Options, Journal of International Money and Finance, 6 (No. 2,
1987), pp. 131152.
Ito, Takatoshi, Foreign Exchange Rate Expectations: Micro Survey Data,
American Economic Review, 80 (No. 3, 1990), pp. 434449.
Ito, Takatoshi, Robert F. Engle, and Wen-Ling Lin, Where Does the Meteor
Shower Come From? The Role of Stochastic Policy Coordination, Journal
of International Economics, 32 (No. 34, 1992), pp. 221240.
Ito, Takatoshi, Richard Lyons, and Michael T. Melvin, Is There Private
Information in the FX Market? The Tokyo Experiment, Journal of Finance,
53 (No. 3, 1998), pp. 11111130.
Jorion, Philippe, Predicting Volatility in the Foreign Exchange Market,
Journal of Finance, 50 (No. 2, 1995), pp. 507528.
, Risk and Turnover in the Foreign Exchange Market, in Frankel, Galli,
and Giovannini, eds., The Microstructure of Foreign Exchange Markets, 1996,
pp. 1937.
Kaminsky, Graciela, and Rodrigo Peruga, Can a Time-Varying Risk Premium
Explain Excess Returns in the Forward Market for Foreign Exchange?
Journal of International Economics, 28 (No. 12, 1990), pp. 4770.
Karpoff, Jonathan M., The Relation between Price Changes and Trading
Volume: A Survey, Journal of Financial and Quantitative Analysis, 22 (No.
1, 1987), pp. 109126.
Kirman, Alan P., Epidemics of Opinion and Speculative Bubbles in Financial
Markets, in Mark P. Taylor, ed., Money and Financial Markets, Oxford,
Blackwell, 1991, pp. 354368.
Kugler, Peter, and Carlos Lenz, Sind Wechselkursfluktuationen zufullig oder
chaotisch? (Are Exchange Rate Fluctuations Random or Chaotic?), Schweizerische Zeitschrift fr Volkswirtschaft und Statistik, 126 (No. 2, 1990), pp.
113128.
Kyle, Albert S., Continuous Auctions and Insider Trading, Econometrica, 53
(No. 6, 1985), pp. 13151335.
Lamoureux, Christopher G., and William D. Lastrapes, Forecasting Stock-Return Variance: Towards an Understanding of Stochastic Implied Volatilities,
Review of Financial Studies, 6 (No. 2, 1993), pp. 293326.
Lane, Philip R., The New Open Economy Macroeconomics: A Survey, CEPR
Discussion Paper No. 2115, London, Centre for Economic Policy Research,
March 1999.
Lee, Tae Hwy, Spread and Volatility in Spot and Forward Exchange Rates,
Journal of International Money and Finance, 13 (No. 3, 1994), pp. 375383.
Lee, Tom K., Does Conditional Covariance or Conditional Variance Explain
Time Varying Risk Premia in Foreign Exchange Returns? Economics Letters,
27 (No. 4, 1988), pp. 371373.

51

Lui, Yu-Hon, and David Mole, The Use of Fundamental and Technical
Analyses by Foreign Exchange Dealers: Hong Kong Evidence, Journal of
International Money and Finance, 17 (No. 3, 1998), pp. 535545.
Lyons, Richard K., Private Beliefs and Information Externalities in the Foreign
Exchange Market, National Bureau of Economic Research Working Paper
No. 3889, Cambridge, Mass., National Bureau of Economic Research, 1991.
, Tests of Microstructural Hypotheses in the Foreign Exchange Market,
Journal of Financial Economics, 39 (No. 23, 1995), pp. 321351.
, Foreign Exchange Volume: Sound and Fury Signifying Nothing? in
Frankel, Galli, and Giovannini, eds., The Microstructure of Foreign Exchange
Markets, 1996, pp. 183201.
, A Simultaneous Trade Model of the Foreign Exchange Hot Potato,
Journal of International Economics, 42 (No. 34, 1997), pp. 275298.
, Profits and Position Control: A Week of FX Dealing, Journal of
International Money and Finance, 17 (No. 1, 1998), pp. 97115.
, The Microstructure Approach to Exchange Rates, Cambridge, Mass.,
MIT Press, forthcoming 2001.
Madhavan, Ananth, and Seymour Smidt, A Bayesian Model of Intraday
Specialist Pricing, Journal of Financial Economics, 30 (No. 1, 1991), pp.
99134.
Malkiel, Burton G., A Random Walk Down Wall Street: Including a Life-Cycle
Guide to Personal Investing, 6th ed., New York and London, Norton, 1996.
McCurdy, Thomas H., and Ieuan G. Morgan, Testing the Martingale Hypothesis in Deutsche Mark Futures with Models Specifying the Form of Heteroscedasticity, Journal of Applied Econometrics, 3 (No. 3, 1988), pp. 187202.
McFarland, James W., R. Richardon Pettit, and Sam K. Sung, The Distribution
of Foreign Exchange Price Changes: Trading Day Effects and Risk Measurement, Journal of Finance, 37 (No. 3, 1982), pp. 693715.
Meese, Richard A., and Andrew K. Rose, An Empirical Assessment of
Nonlinearities in Models of Exchange Rate Determination, Review of
Economic Studies, 58 (No. 3, 1991), pp. 603619.
Melvin, Michael T., and Xixi Yin, Public Information Arrival, Exchange Rate
Volatility, and Quote Frequency, Department of Economics, Arizona State
University, Tempe, Ariz., 1999, processed.
Melvin, Michael T., and Patricia Ramirez, FOMC Days and Exchange Rate
Dynamics, Department of Economics, Arizona State University, Tempe,
Ariz., 2000, processed.
Menkhoff, Lukas, Examining the Use of Technical Currency Analysis,
International Journal of Finance and Economics, 2 (No. 4, 1997), pp. 307318.
, The Noise Trading ApproachQuestionnaire Evidence from Foreign
Exchange, Journal of International Money and Finance, 17 (No. 3, 1998),
pp. 547564.
Milhoj, Anders, A Conditional Variance Model for Daily Deviations of an
Exchange Rate, Journal of Business and Economic Statistics, 5 (No. 1,
1987), pp. 99103.

52

Obstfeld, Maurice, and Kenneth Rogoff, Exchange Rate Dynamics Redux,

Journal of Political Economy, 103 (No. 3, 1995), pp. 624660.
, Foundations of International Macroeconomics, Cambridge, Mass., MIT
Press, 1996.
Oldfield, George S., Jr., and Richard J. Rogalski, A Theory of Common Stock
Returns over Trading and Non-Trading Periods, Journal of Finance, 35 (No.
3, 1980), pp. 729751.
Overturf, Stephen Frank, Risk, Transactions Charges, and the Market for
Foreign Exchange Services, Economic Inquiry, 20 (No. 2, 1982), pp.
291302.
Perraudin, William, and Paolo Vitale, Interdealer Trade and Information Flows
in a Decentralized Foreign Exchange Market, in Frankel, Galli, and
Giovannini, eds., The Microstructure of Foreign Exchange Markets, 1996, pp.
7399.
Pesaran, M. Hashem, Consistency of Short-Term and Long-Term Expectations, Journal of International Money and Finance, 8 (No. 4, 1989), pp.
511516.
Richardson, Matthew, and Tom Smith, A Direct Test of the Mixture of
Distributions Hypothesis: Measuring the Daily Flow of Information, Journal
of Financial and Quantitative Analysis, 29 (No. 1, 1994), pp. 101116.
Sarno, Lucio, Towards a New Paradigm in Open Economy Modeling: Where
Do We Stand? Federal Reserve Bank of St. Louis Review, forthcoming 2001.
Schwartz, Robert A., Equity Markets: Structure, Trading, and Performance,
New York, Harper and Row, 1988.
Scott, Louis O., Option Pricing When the Variance Changes Randomly:
Theory, Estimation, and an Application, Journal of Financial and Quantitative Analysis, 22 (No. 4, 1987), pp. 419438.
Shiller, Robert J., Stock Prices and Social Dynamics, Brookings Papers on
Economic Activity, 0 (No. 2, 1984), pp. 457498; reprinted in Shiller, Market
Volatility, Cambridge, Mass., MIT Press, 1989, pp. 748.
Smidt, Seymour, Which Road to an Efficient Stock Market, Financial Analysts
Journal, 22 (No. 1, 1971), pp. 6469.
Stigler, George J., Public Regulation of the Securities Markets, Journal of
Business, 37 (No. 4, 1964), pp. 11711142.
Stein, Jeremy C., Informational Externalities and Welfare-Reducing Speculation, Journal of Political Economy, 95 (No. 6, 1987), pp. 11231145.
Stoll, Hans R., The Supply of Dealer Services in Securities Markets, Journal
of Finance, 33 (No. 4, 1978), pp. 11331151.
Subrahmanyam, Avanidhar, Risk Aversion, Market Liquidity, and Price
Efficiency, Review of Financial Studies, 4 (No. 3, 1991), pp. 417441.
Suvanto, Antti, Foreign Exchange Dealing: Essays on the Microstructure of the
Foreign Exchange Market, Sarja A 19 Series, Research Institute of the
Finnish Economy, Helsinki, 1993.
Takagi, Shinji, Exchange Rate Expectations: A Survey of Survey Studies,
International Monetary Fund Staff Papers, 38 (No. 1, 1991), pp. 156183.

53

Tauchen, George E., and Mark Pitts, The Price Variability-Volume Relationship on Speculative Markets, Econometrica, 51 (No. 2, 1983), pp. 485505.
Taylor, Mark P., Covered Interest Parity: A High-Frequency, High-Quality
Data Study, Economica, 54 (No. 216, 1987), pp. 429438.
, Covered Interest Arbitrage and Market Turbulence, Economic Journal,
99 (No. 396, 1989), pp. 376391.
, The Economics of Exchange Rates, Journal of Economic Literature,
83 (No. 1, 1995), pp. 1347.
Taylor, Mark P., and Helen Allen, The Use of Technical Analysis in the
Foreign Exchange Market, Journal of International Money and Finance, 11
(No. 3, 1992), pp. 304314.
Wasserfallen, Walter, and Heinz Zimmermann, The Behavior of Intra-Daily
Exchange Rates, Journal of Banking and Finance, 9 (No. 1, 1985), pp.
5572.
Wei, Shang-Jin, and Jeffrey A. Frankel, Are Option-Implied Forecasts of
Exchange Rate Volatility Excessively Variable? National Bureau of Economic
Research Working Paper No. 3910, Cambridge, Mass., National Bureau of
Economic Research, 1991.
Westerfield, Randolph, The Distribution of Common Stock Price Changes: An
Application of Transactions Time and Subordinated Stochastic Models,
Journal of Financial and Quantitative Analysis, 12 (No. 5, 1977), pp.
743765.
Wiggins, James B., Option Values under Stochastic Volatility: Theory and
Empirical Estimates, Journal of Financial Economics, 19 (No. 2, 1987), pp.
351372.
Wolinsky, Asher, Information Revelation in a Market with Pairwise Meetings,
Econometrica, 58 (No. 1, 1990), pp. 123.

54

PUBLICATIONS OF THE
INTERNATIONAL ECONOMICS SECTION
Notice to Contributors
The International Economics Section publishes papers in two series. ESSAYS IN
INTERNATIONAL ECONOMICS and PRINCETON STUDIES IN INTERNATIONAL ECONOMICS. Two earlier series, REPRINTS IN INTERNATIONAL FINANCE and SPECIAL
PAPERS IN INTERNATIONAL ECONOMICS, have been discontinued, with the SPECIAL
PAPERS being absorbed into the STUDIES series.
The Section welcomes the submission of manuscripts focused on topics in international trade, international macroeconomics, or international finance. Submissions
should address systemic issues for the global economy or, if concentrating on
particular economies, should adopt a comparative perspective.
ESSAYS IN INTERNATIONAL ECONOMICS are meant to disseminate new views about
international economic events and policy issues. They should be accessible to a broad
audience of professional economists.
PRINCETON STUDIES IN INTERNATIONAL ECONOMICS are devoted to new research
in international economics or to synthetic treatments of a body of literature. They
should be comparable in originality and technical proficiency to papers published in
leading economic journals. Papers that are longer and more complete than those
publishable in the professional journals are welcome.
Manuscripts should be submitted in triplicate, typed single sided and double
spaced throughout on 8 by 11 white bond paper. Publication can be expedited if
manuscripts are computer keyboarded in WordPerfect or a compatible program.
Additional instructions and a style guide are available from the Section or on the
website at www.princeton.edu/~ies.
How to Obtain Publications
The Sections publications are distributed free of charge to college, university, and
public libraries and to nongovernmental, nonprofit research institutions. Eligible
institutions may ask to be placed on the Sections permanent mailing list.
Individuals and institutions not qualifying for free distribution may receive all
publications for the calendar year for a subscription fee of $45.00. Late subscribers
will receive all back issues for the year during which they subscribe.
Publications may be ordered individually, with payment made in advance. All
publications (ESSAYS, STUDIES, SPECIAL PAPERS, and REPRINTS) cost $10.00 each;
an additional $1.50 should be sent for postage and handling within the United States,
Canada, and Mexico; $4 should be added for surface delivery outside the region.
All payments must be made in U.S. dollars. Subscription fees and charges for
single issues will be waived for organizations and individuals in countries where
foreign-exchange regulations prohibit dollar payments.
Information about the Section and its publishing program is available on the
Sections website at www.princeton.edu/~ies. A subscription and order form is
printed at the end of this volume. Correspondence should be addressed to:
International Economics Section
Department of Economics, Fisher Hall
Princeton University
Princeton, New Jersey 08544-1021
Tel: 609-258-4048 Fax: 609-258-1374
E-mail: [email protected]

55

List of Recent Publications

A complete list of publications is available at the International Economics Section
website at www.princeton.edu/~ies.
ESSAYS IN INTERNATIONAL ECONOMICS

(formerly Essays in International Finance)

183. Michael Bruno, High Inflation and the Nominal Anchors of an Open Economy.
(June 1991)
184. Jacques J. Polak, The Changing Nature of IMF Conditionality. (September
1991)
185. Ethan B. Kapstein, Supervising International Banks: Origins and Implications
of the Basle Accord. (December 1991)
186. Alessandro Giustiniani, Francesco Papadia, and Daniela Porciani, Growth and
Catch-Up in Central and Eastern Europe: Macroeconomic Effects on Western
Countries. (April 1992)
187. Michele Fratianni, Jrgen von Hagen, and Christopher Waller, The Maastricht
Way to EMU. (June 1992)
188. Pierre-Richard Agnor, Parallel Currency Markets in Developing Countries:
Theory, Evidence, and Policy Implications. (November 1992)
189. Beatriz Armendariz de Aghion and John Williamson, The G-7s Joint-andSeveral Blunder. (April 1993)
190. Paul Krugman, What Do We Need to Know about the International Monetary
System? (July 1993)
191. Peter M. Garber and Michael G. Spencer, The Dissolution of the AustroHungarian Empire: Lessons for Currency Reform. (February 1994)
192. Raymond F. Mikesell, The Bretton Woods Debates: A Memoir. (March 1994)
193. Graham Bird, Economic Assistance to Low-Income Countries: Should the Link
be Resurrected? (July 1994)
194. Lorenzo Bini-Smaghi, Tommaso Padoa-Schioppa, and Francesco Papadia, The
Transition to EMU in the Maastricht Treaty. (November 1994)
195. Ariel Buira, Reflections on the International Monetary System. (January 1995)
196. Shinji Takagi, From Recipient to Donor: Japans Official Aid Flows, 1945 to
1990 and Beyond. (March 1995)
197. Patrick Conway, Currency Proliferation: The Monetary Legacy of the Soviet
Union. (June 1995)
198. Barry Eichengreen, A More Perfect Union? The Logic of Economic Integration.
(June 1996)
199. Peter B. Kenen, ed., with John Arrowsmith, Paul De Grauwe, Charles A. E.
Goodhart, Daniel Gros, Luigi Spaventa, and Niels Thygesen, Making EMU
HappenProblems and Proposals: A Symposium. (August 1996)
200. Peter B. Kenen, ed., with Lawrence H. Summers, William R. Cline, Barry
Eichengreen, Richard Portes, Arminio Fraga, and Morris Goldstein, From Halifax
to Lyons: What Has Been Done about Crisis Management? (October 1996)
201. Louis W. Pauly, The League of Nations and the Foreshadowing of the International Monetary Fund. (December 1996)

56

202. Harold James, Monetary and Fiscal Unification in Nineteenth-Century Germany:

What Can Kohl Learn from Bismarck? (March 1997)
203. Andrew Crockett, The Theory and Practice of Financial Stability. (April 1997)
204. Benjamin J. Cohen, The Financial Support Fund of the OECD: A Failed
Initiative. (June 1997)
205. Robert N. McCauley, The Euro and the Dollar. (November 1997)
206. Thomas Laubach and Adam S. Posen, Disciplined Discretion: Monetary Targeting in Germany and Switzerland. (December 1997)
207. Stanley Fischer, Richard N. Cooper, Rudiger Dornbusch, Peter M. Garber,
Carlos Massad, Jacques J. Polak, Dani Rodrik, and Savak S. Tarapore, Should
the IMF Pursue Capital-Account Convertibility? (May 1998)
208. Charles P. Kindleberger, Economic and Financial Crises and Transformations
in Sixteenth-Century Europe. (June 1998)
209. Maurice Obstfeld, EMU: Ready or Not? (July 1998)
210. Wilfred Ethier, The International Commercial System. (September 1998)
211. John Williamson and Molly Mahar, A Survey of Financial Liberalization.
(November 1998)
212. Ariel Buira, An Alternative Approach to Financial Crises. (February 1999)
213. Barry Eichengreen, Paul Masson, Miguel Savastano, and Sunil Sharma,
Transition Strategies and Nominal Anchors on the Road to Greater ExchangeRate Flexibility. (April 1999)
214. Curzio Giannini, Enemy of None but a Common Friend of All? An International Perspective on the Lender-of-Last-Resort Function. (June 1999)
215. Jeffrey A. Frankel, No Single Currency Regime Is Right for All Countries or at
All Times. (August 1999)
216. Jacques J. Polak, Streamlining the Financial Structure of the International
Monetary Fund. (September 1999)
217. Gustavo H. B. Franco, The Real Plan and the Exchange Rate. (April 2000)
218. Thomas D. Willett, International Financial Markets as Sources of Crises or
Discipline: The Too Much, Too Late Hypothesis. (May 2000)
219. Richard H. Clarida, G-3 Exchange-Rate Relationships: A Review of the Record
and of Proposals for Change. (September 2000)
220. Stanley Fischer, On the Need for an International Lender of Last Resort.
(November 2000)
221. Benjamin J. Cohen, Life at the Top: International Currencies in the TwentyFirst Century. (December 2000)
PRINCETON STUDIES IN INTERNATIONAL ECONOMICS

(formerly Princeton Studies in International Finance)

70. Kaushik Basu, The International Debt Problem, Credit Rationing and Loan
Pushing: Theory and Experience. (October 1991)
71. Daniel Gros and Alfred Steinherr, Economic Reform in the Soviet Union: Pas de Deux
between Disintegration and Macroeconomic Destabilization. (November 1991)
72. George M. von Furstenberg and Joseph P. Daniels, Economic Summit Declarations, 1975-1989: Examining the Written Record of International Cooperation. (February 1992)

57

73. Ishac Diwan and Dani Rodrik, External Debt, Adjustment, and Burden Sharing:
A Unified Framework. (November 1992)
74. Barry Eichengreen, Should the Maastricht Treaty Be Saved? (December 1992)
75. Adam Klug, The German Buybacks, 1932-1939: A Cure for Overhang?
(November 1993)
76. Tamim Bayoumi and Barry Eichengreen, One Money or Many? Analyzing the
Prospects for Monetary Unification in Various Parts of the World. (September 1994)
77. Edward E. Leamer, The Heckscher-Ohlin Model in Theory and Practice.
(February 1995)
78. Thorvaldur Gylfason, The Macroeconomics of European Agriculture. (May 1995)
79. Angus S. Deaton and Ronald I. Miller, International Commodity Prices, Macroeconomic Performance, and Politics in Sub-Saharan Africa. (December 1995)
80. Chander Kant, Foreign Direct Investment and Capital Flight. (April 1996)
81. Gian Maria Milesi-Ferretti and Assaf Razin, Current-Account Sustainability.
(October 1996)
82. Pierre-Richard Agnor, Capital-Market Imperfections and the Macroeconomic
Dynamics of Small Indebted Economies. (June 1997)
83. Michael Bowe and James W. Dean, Has the Market Solved the Sovereign-Debt
Crisis? (August 1997)
84. Willem H. Buiter, Giancarlo M. Corsetti, and Paolo A. Pesenti, Interpreting the
ERM Crisis: Country-Specific and Systemic Issues. (March 1998)
85. Holger C. Wolf, Transition Strategies: Choices and Outcomes. (June 1999)
86. Alessandro Prati and Garry J. Schinasi, Financial Stability in European Economic
and Monetary Union. (August 1999)
87. Peter Hooper, Karen Johnson, and Jaime Marquez, Trade Elasticities for the
G-7 Countries. (August 2000)
88. Ramkishen S. Rajan, (Ir)relevance of Currency-Crisis Theory to the Devaluation
and Collapse of the Thai Baht. (February 2001)
89. Lucio Sarno and Mark P. Taylor, The Microstructure of the Foreign-Exchange
Market: A Selective Survey of the Literature. (May 2001)
SPECIAL PAPERS IN INTERNATIONAL ECONOMICS

16. Elhanan Helpman, Monopolistic Competition in Trade Theory. (June 1990)

17. Richard Pomfret, International Trade Policy with Imperfect Competition. (August 1992)
18. Hali J. Edison, The Effectiveness of Central-Bank Intervention: A Survey of the
Literature After 1982. (July 1993)
19. Sylvester W.C. Eijffinger and Jakob De Haan, The Political Economy of CentralBank Independence. (May 1996)
20. Olivier Jeanne, Currency Crises: A Perspective on Recent Theoretical Developments. (March 2000)
REPRINTS IN INTERNATIONAL FINANCE

29. Peter B. Kenen, Sorting Out Some EMU Issues; reprinted from Jean Monnet
Chair Paper 38, Robert Schuman Centre, European University Institute, 1996.
(December 1996)

58

The work of the International Economics Section is supported

in part by the income of the Walker Foundation, established
in memory of James Theodore Walker, Class of 1927. The
offices of the Section, in Fisher Hall, were provided by a
generous grant from Merrill Lynch & Company.

ISBN 0-88165-261-X
Recycled Paper