Electronic copy available at: http://ssrn.com/abstract=1727410
Michael R King
The $4 trillion question: what explains FX growth
since the 2007 survey?1
Daily average foreign exchange market turnover reached $4 trillion in April 2010, 20%
higher than in 2007. Growth owed largely to the increased trading activity of “other
financial institutions”, which contributed 85% of the higher turnover. Within this
customer category, the growth is driven by high-frequency traders, banks trading as
clients of the biggest dealers, and online trading by retail investors. Electronic trading
has been instrumental to this increase, particularly algorithmic trading.
JEL Classification: F31, G12, G15, C42, C82.
In April this year, 53 central banks and monetary authorities participated in the
eighth Triennial Central Bank Survey of Foreign Exchange and Derivatives
Market Activity (“the Triennial”).2 The 2010 Triennial shows a 20% increase in
global foreign exchange (FX) market activity over the past three years, bringing
average daily turnover to $4.0 trillion (Graph 1, left-hand panel).3 While the
growth in FX market activity since 2007 is substantial, it represents a slowdown
following the unprecedented 72% rise between 2004 and 2007.4 However,
against the backdrop of the global financial crisis of 2007–09 and the recent
turmoil in European sovereign bond markets, the continued growth
demonstrates the resilience of this market.
1 The authors would like to thank Claudio Borio, Alain Chaboud, Liz Costin, Gabriele Galati,
Simon Jones, Colin Lambert, Angelo Ranaldo, Elvira Sojli, Christian Upper and Mark Warms
for useful comments and suggestions. Jakub Demski, Victoria Halstensen, Carlos Mallo and
Jhuvesh Sobrun provided excellent research assistance. Data were generously provided by
the Chicago Mercantile Exchange, and ICAP EBS. The authors are grateful to representatives
from five single-bank and 12 multi-bank trading systems who agreed to be interviewed and
shared proprietary data on their activities. The views expressed in this article are those of the
authors and do not necessarily reflect those of the BIS or the Central Bank of Norway.
2 The Triennial has been conducted every three years since 1989. For details on the
methodology and changes over time, see the article by King and Mallo (this issue). For more
details on developments in emerging market currencies, see Mihaljek and Packer (this issue).
Detailled results of the 2010 survey are available at www.bis.org/publ/rpfxf10t.htm.
3 Because euro/dollar exchange rates were very similar in April 2007 and 2010, growth
calculated at constant exchange rates was also similar, at 18%.
4 For more details on the results of the 2007 Triennial, see Galati and Heath (2007).
BIS Quarterly Review, December 2010 27
Electronic copy available at: http://ssrn.com/abstract=1727410
The 2010 Triennial data show that 85% of the growth in FX market
turnover since 2007 reflects the increased trading activity of “other financial
institutions” (Table 1). This broad category includes smaller banks, mutual
funds, money market funds, insurance companies, pension funds, hedge funds,
currency funds and central banks, among others. For the first time, activity by
other financial institutions surpassed transactions between reporting dealers
(ie inter-dealer trades), reflecting a trend that has been evident over the past
decade (Graph 1, centre panel).5
While FX turnover grew by 20% between April 2007 and 2010, trading by
corporations and governments fell by 10% over this period, possibly reflecting
slower economic growth (see box “Foreign exchange turnover versus
international trade and financial flows”). The reduced FX activity by
corporations is mirrored in the decline in international banking activity,
particularly syndicated loans, and in trade finance (Chui et al (2010)).
Given that most of the growth in FX market activity since 2007 is due to
increased trading by other financial institutions, the $4 trillion dollar question is:
which financial institutions are behind this growth? The Triennial data do not
break down trades within this category of counterparty. Discussions with
market participants, data from regional FX surveys and an analysis of the
currency composition and location of trading activity provide some useful clues.
Taken together, they suggest the increased turnover is driven by: (i) greater
activity of high-frequency traders; (ii) more trading by smaller banks that are
increasingly becoming clients of the top dealers for the major currency pairs;
and (iii) the emergence of retail investors (both individuals and smaller
Increase in global FX market turnover by counterparty1
to FX market
Global FX market 3,981 657 20 100
Reporting dealers 1,548 156 11 24
Other financial institutions 1,900 561 42 85
Non-financial customers 533 –60 –10 –9
Spot 1,490 485 48 74
Outright forwards 475 113 31 17
FX swaps 1,765 51 3 8
Currency options 207 –4 –2 –1
Currency swaps 43 11 36 2
1 Adjusted for local and cross-border double-counting, ie “net-net” basis.
2 In billions of US dollars.
3 Percentage contribution to the total increase of $657 billion from 2007 to 2010.
Source: 2010 Triennial Central Bank Survey. Table 1
5 A glossary at the end explains the italicised terms that appear in this feature.
28 BIS Quarterly Review, December 2010
Electronic copy available at: http://ssrn.com/abstract=1727410
Global foreign exchange market turnover
Daily averages in April
By counterparty1, 2 Share of total turnover by
counterparty1, 3 Number of banks accounting for
75% of traditional FX turnover4
2001 2004 2007 2010
Other financial institutions
institutions) as a significant category of FX market participants.6 This article
explores the contribution of each of these customer types to the growth of
global FX turnover.
An important structural change enabling increased FX trading by these
customers is the spread of electronic execution methods. Electronic trading
and electronic brokering are transforming FX markets by reducing transaction
costs and increasing market liquidity. These changes, in turn, are encouraging
greater participation across different customer types.
Continued investment in electronic execution methods has paved the way
for the growth of algorithmic trading. In algorithmic trading, investors connect
their computers directly with trading systems known as electronic
communication networks (ECNs). Examples of ECNs in FX markets are
electronic broking systems (such as EBS and Thomson Reuters Matching),
multi-bank trading systems (such as Currenex, FXall and Hotspot FX) and
single-bank trading systems. A computer algorithm then monitors price quotes
collected from different ECNs and places orders without human intervention
(Chaboud et al (2009)). High-frequency trading (HFT) is one algorithmic
strategy that profits from incremental price movements with frequent, small
trades executed in milliseconds.
While banks engaged in FX markets below the top tier continue to be
important players, the long-term trend towards greater concentration of FX
activity in a few global banks continues (Graph 1, right-hand panel). The
largest dealers have seen their FX business grow by investing heavily in their
single-bank proprietary trading systems. The tight bid-ask spreads and
guaranteed market liquidity on such platforms are making it unprofitable for
smaller players to compete for customers in the major currency pairs.
6 Trading by all three of these groups is categorised by reporting dealers as trades with other
financial institutions. Retail trades are routed through online platforms that are classified as
other financial institutions by reporting dealers.
1995 1998 2001 2004 2007 2010
Hong Kong SAR
2001 2004 2007 2010
1 Adjusted for local and cross-border inter-dealer double-counting, ie “net-net” basis.
2 In billions of US dollars.
3 In per
cent. 4 Relates to spot, outright forwards and FX swaps.
Source: Triennial Central Bank Survey. Graph 1
… enabled by
electronic trading …
… with banks
trading as clients of
top dealers …
BIS Quarterly Review, December 2010 29
Foreign exchange turnover versus international trade and financial flows
The FX market is the largest financial market in the world, but how does turnover in this market compare
with real activity? This box compares turnover in the seven countries that have the most active FX
markets with the level of GDP and the volume of trade. It also benchmarks FX activity to trading volumes
on major stock exchanges. The seven countries with the most active FX markets are (in decreasing
order): the United Kingdom, the United States, Japan, Singapore, Switzerland, Hong Kong SAR and
The motives for trading a currency may be divided into transactions linked to cross-border
trade in goods and services, and transactions related to cross-border financial flows. The left-hand
panel of Graph A shows the ratio of global FX turnover for a country, compared to the country’s
GDP. FX turnover is several times larger than the total output of the economy. The FX
turnover/GDP ratio is smallest for the largest economies, the United States and Japan. In these two
countries, FX turnover is more than 14 times GDP. In most cases, FX turnover has grown faster
than GDP, as indicated by the upward-sloping lines.
Graph A also looks at the FX trading activity of different customer types. The centre panel
compares FX turnover by other financial institutions with activity on a country’s stock exchanges.
“Other financial institutions” is a broad category that includes asset managers and institutional
investors, who are most likely to be active in cross-border financial markets. (While it would be
more appropriate to compare FX turnover with trading volumes in bond markets, where FX hedging
activity is more prevalent, data on bond turnover are not available.) FX market turnover is many
times larger than equity trading volumes. Again, the ratio of FX turnover to equity turnover is
smallest for the United States and Japan, but still sizeable. The growth in FX turnover since 2007 is
much stronger than that of equity trading for several countries, as seen in the sharp increase in the
Foreign exchange market turnover and rationale for trading
In per cent
Total FX turnover / GDP1 FX turnover by other financial
institutions / equity turnover FX turnover by non-financial
customers / imports + exports
92 95 98 01 04 07 10
Hong Kong SAR
United Kingdom 40
92 95 98 01 04 07 10 92 95 98 01 04 07 10
1 Foreign exchange turnover adjusted for local inter-dealer double-counting, ie “net-gross” basis.
Sources: IMF IFS; World Federation of Exchanges; BIS. Graph A
Finally, the right-hand panel of Graph A shows the ratio of FX activity by non-financial
customers to gross trade flows. Gross trade flows are defined as the sum of imports and exports of
goods and services. FX turnover is much higher than underlying trade flows, although the ratios are
an order of magnitude smaller than in the other two panels. A closer look at growth since 2007
shows the decrease in activity by non-financial customers is matched by a drop in trade volumes, at
least for the United States and the United Kingdom.
Overall, looking at developments since 1992, it is clear that FX turnover has increased more
than underlying economic activity, whether measured by GDP, equity turnover or gross trade flows.
All comparisons are based on monthly figures, where daily average FX turnover is multiplied by 20 trading days,
and measures of economic activity are yearly figures divided by 12.
30 BIS Quarterly Review, December 2010
Increasingly, many smaller banks are becoming clients of the top dealers
for these currencies, while continuing to make markets for customers in local
currencies. This hybrid role allows smaller banks with client relationships to
profit from their local expertise and comparative advantage in the provision of
credit, while freeing them from the heavy investment required to compete in
spot market-making for the major currency pairs.
Finally, greater FX trading activity by small retail investors has made a
significant contribution to growth in spot FX, and this growth in activity was
made possible by the spread of electronic execution methods.
… and individual
The 2007–09 financial crisis and its impact on FX markets
While the $4 trillion figure reported in the 2010 Triennial sets a new record high
for daily average FX turnover, this level may already have been reached
18 months ago during the 2007–09 global financial crisis. Data from regional
foreign exchange committees and multi-bank ECNs show a peak in FX activity
in October 2008 following the bankruptcy of Lehman Brothers (Graph 2, left-
hand panel). Thereafter, activity in FX markets declined sharply, before
recovering from October 2009 onwards.
Global FX activity
peaked in October
Following Lehman’s bankruptcy, many financial markets experienced large
disruptions with a sharp increase in volatility (Graph 2, centre panel). With
limited market liquidity in various asset classes, many investors reportedly
turned to spot FX markets to hedge risk exposures (“proxy hedging”). For
example, downside risk in US equities was reportedly hedged by buying
Japanese yen, in European equities by selling the euro, and in emerging
market equities using emerging market currencies. These strategies may have
had limited success, but at least they were available – albeit at an increased
cost, as bid-ask spreads for the major currencies widened during the height of
FX markets during recent crises
Spot FX turnover from regional
surveys1 Implied volatility for main currency
pairs2, 3 Barclays’ FX Liquidity Index3, 4
2007 2008 2009 2010
2006 2007 2008 2009 2010
2008 2009 2010
1 Daily averages based on semi annual surveys; in billions of US dollars. 2 One-month at-the-money implied volatilities based on
options. 3 Vertical lines represent Lehman Brothers’ bankruptcy (15 September 2008) and the S&P downgrade of Greek government
debt (27 April 2010); in percentage points. 4 A higher index reflects better liquidity conditions. The FX Liquidity Index is constructed
using the notional amounts traded for a fixed set of FX spreads, aggregated using a weighting by currency pair.
Sources: Australian Foreign Exchange Committee; Barclays Capital; Bloomberg; Canadian Foreign Exchange Committee; The London
Foreign Exchange Joint Standing Committee; The New York Foreign Exchange Committee; Singapore Foreign Exchange Market
Committee. Graph 2
… as investors
used FX to hedge
BIS Quarterly Review, December 2010 31
the crisis by a factor of 4 to 5 times (Melvin and Taylor (2009)) A proprietary
liquidity index constructed by Barclays shows that market liquidity for spot
trading in major currency pairs dropped sharply around this event, as well as
following the downgrade of Greek government debt on 27 April 2010 (Graph 2,
The rise in FX volatility and increased risk aversion of investors led to a
rapid unwind of currency carry trade positions, with funding currencies
appreciating sharply and many investors experiencing large losses. The
Japanese yen, for example, appreciated by 7.7% against the Australian dollar
on 16 August 2008. While the unwinding of carry trades may have been
important over 2008 and 2009, market participants report that this was not a
significant factor explaining FX turnover during April 2010.
Despite the widespread financial market disruptions, most parts of the FX
markets continued to function relatively smoothly, although FX swaps were
severely disrupted. The robustness of FX markets in the face of these
disruptions owes much to the role of CLS Bank, which uses a combination of
payment versus payment in central bank funds and multilateral payment netting
to eliminate settlement risk (Galati (2002), Lindley (2008)). In the aftermath of
the global financial crisis, CLS Bank has seen an influx of new members,
particularly investment and pension funds.7
FX markets proved
robust due to CLS
The global financial crisis has changed the focus in FX markets and
attracted the attention of regulators. Clients are concerned about minimising
transaction costs while demonstrating best execution. Managing counterparty
credit risk, while always important in FX markets, has taken on increased
importance. Activity in instruments that generate counterparty credit risk
exposures, such as FX swaps, has not rebounded due to continuing constraints
on dealers’ balance sheets and restrictions on the availability of credit.
Customers are reportedly relying more on bank credit lines and central bank
facilities instead of the FX swap market. Regulators have increased capital
requirements for retail FX brokers and reduced the leverage available to
individuals. Finally, regulators are focused on reducing systemic risk and
increasing the robustness of electronic infrastructure by increasing the use of
risk is more
Electronic execution methods are transforming the FX market
The greater activity of all three of the above-mentioned customer types – high-
frequency traders, banks as clients and retail investors – is closely related to
the growth of electronic execution methods in FX markets. Greenwich
Associates estimates that more than 50% of total foreign exchange trading
7 Based on the Triennial data for April 2010, CLS Bank settled 43% of spot transactions and
39% of combined spot, outright forwards and FX swaps (compared to 42% and 34% in April
8 In the United States, the 2010 Dodd-Frank Act will make central clearing of OTC derivatives
mandatory for many investors. In Europe, trading of OTC derivatives is being addressed in the
review of the Markets in Financial Instruments Directive (MiFID). For more on central
counterparties in OTC markets, see Cecchetti et al (2009).
32 BIS Quarterly Review, December 2010
volume is now being executed electronically (Graph 3, left-hand panel).
Electronic execution methods can be divided into three categories: electronic
brokers, multi-bank trading systems and single-bank trading systems.
Electronic brokers were introduced in the inter-dealer FX market as early
as in 1992. For customers, however, the main channel for trading continued to
be direct contact with dealers by telephone. In the rather opaque and
fragmented FX market of the 1990s, barriers to entry were high and
competition was limited. Customers typically paid large spreads on their FX
arrived in 1992 …
The first multi-bank trading system was Currenex, which was launched in
1999. By providing customers with competing quotes from different FX dealers
on a single page, Currenex increased transparency, reduced transaction costs
and attracted a growing customer base. State Street’s FXConnect, which had
been launched in 1996 as a single-bank trading system servicing only State
Street’s clients, opened up in 2000 and became a multi-bank ECN.
… followed by
multi-bank ECNs in
the late 1990s ...
In response to the increased competition, top FX dealers launched
proprietary single-bank trading systems for their clients, such as Barclays’
BARX in 2001, Deutsche Bank’s Autobahn in 2002 and Citigroup’s Velocity in
2006. According to data provided to the BIS, daily average trading volumes on
the top single-bank trading systems have increased by up to 200% over the
past three years.
Structural changes in execution methods are moving fastest in the largest
financial centres. Table 2 shows the execution methods across all FX
instruments as reported in the 2010 Triennial. According to the survey
methodology, each country allocates all OTC FX transactions to one of these
categories.9 The table compares the top three financial centres (the United
Kingdom, the United States and Japan) with the next seven most active
Growth of algorithmic trading
Share of total FX volume traded
electronically1, 2 Share of manual vs algorithmic
trading on EBS1 Average daily FX turnover on CME
2004 2007 2010
98% 72% 55%
Americas Europe Japan 00 02 04 06 08 10
1 In per cent. 2 Based on Greenwich Associates survey of top corporations and financial institutions.
3 Notional amounts, in
billions of US dollars.
Sources: CME; EBS; Greenwich Associates. Graph 3
…with the biggest
impact in financial
9 For more details on the categories of execution methods, see King and Mallo (this issue).
BIS Quarterly Review, December 2010 33
Execution methods for global FX market turnover1
In per cent
All FX instruments
UK, US and Japan 15 16 19 39 12 100
Next 7 countries4 24 19 17 31 10 100
Remaining 43 countries 29 11 24 25 10 100
UK, US and Japan 12 8 27 36 16 100
Next 7 countries4 20 11 20 39 11 100
Remaining 43 countries 27 8 27 28 10 100
FX swaps only
UK, US and Japan 18 28 15 32 7 100
Next 7 countries4 26 23 17 25 9 100
Remaining 43 countries 33 15 22 18 12 100
1 When comparing national results, FX turnover is on a “net-gross” basis (ie only adjusting for local inter-dealer double-
counting). 2 Trades directly between reporting dealers executed either electronically or by telephone. 3 All direct trades between a
customer and a dealer executed either by telephone or on a single-bank trading system. 4 In descending order of global FX activity:
Switzerland, Singapore, Hong Kong SAR, Australia, France, Denmark and Germany.
Source: 2010 Triennial Central Bank Survey. Table 2
countries in the 2010 Triennial.10 These top 10 countries account for close to
90% of global FX turnover, with volumes dropping off sharply thereafter.
Table 2 also shows the execution methods for the remaining 43 countries in the
In the top three financial centres, customer direct trading – whether
executed electronically on a single-bank portal or by telephone – is the most
important category.11 The growth since 2007 is primarily due to the increased
importance of single-bank trading systems. The share of customer direct
trading (39%) has grown at the expense of inter-dealer direct trading (15%), as
seen in the comparison across the different country groups. Notice that the
relative shares of voice broking and electronic broking are similar for the top
10 financial centres, but electronic broking is much more important than voice
for the remaining 43 countries.
A comparison of the execution method for spot trades and FX swaps
highlights the areas where electronic methods are gaining ground. Electronic
broking and multi-bank trading systems are more important for spot trading,
where counterparty credit and settlement risks are limited. Instruments that
embody counterparty credit risk, such as FX swaps, are harder to trade
electronically. Individual transactions in FX swaps tend to be large and
has driven spot FX
10 The next seven most active countries in the 2010 Triennial, in descending order of FX activity,
are: Switzerland, Singapore, Hong Kong SAR, Australia, France, Denmark and Germany.
11 While “customer direct” and “single-bank trading systems” are separate execution categories
in the Triennial, both categories may include electronic trades. Some reporting dealers appear
to have allocated their electronic trades to the former category while others used the latter.
34 BIS Quarterly Review, December 2010
negotiated on a bilateral basis. A FX swap generates a credit exposure to the
counterparty, particularly on longer-dated instruments. Given the greater risk,
banks want to be able to check their credit limits with counterparties on a more
real-time basis. As a result, in all country groupings, a greater share of FX
swaps is therefore transacted via inter-dealer direct and voice brokers.
The growth of single-bank trading systems has brought several important
changes to the FX market. The biggest FX dealers, such as Barclays,
Deutsche Bank and UBS, have gained market share, reaping the benefits of
their IT investment, while contributing to the overall growth of global FX
markets. In a number of major currency pairs, many smaller banks are
reportedly becoming clients of the top FX dealers. In this hybrid role, smaller
banks may trade the major currencies either via electronic brokers (such as
EBS or Thomson Reuters Matching) or via a top dealer’s single-bank trading
system, while focusing on making markets in their local currency. These
structural changes have increased turnover by other financial institutions and
decreased the relative share of inter-dealer activity.
dealers to launch
own platforms …
The cost-effectiveness of electronic trading and the increased competition
have led to lower transaction costs for customers, in turn supporting turnover.
Table 3 shows the increase in customer direct trading of spot FX, whether
executed by telephone or on single-bank trading systems. Across all countries,
this activity grew by 67% over the past three years, outpacing the overall 50%
growth in spot. When ranked based on the biggest absolute increases,
customer trading of spot in the United Kingdom more than doubled, while
turnover for the United States, Australia, Denmark and Japan also increased
substantially. Emerging markets are also contributing, notably Hong Kong SAR,
Singapore, Brazil and India.
Customer direct trading of spot FX globally1
20102 20072 Change from
United Kingdom 214.8 94.3 120.4 128
United States 168.3 116.9 51.5 44
Japan 43.6 33.8 9.8 29
Australia 32.2 8.0 24.2 303
Denmark 17.4 2.5 14.9 597
Singapore 16.3 14.1 2.2 15
Hong Kong SAR 10.1 5.4 4.7 86
Canada 4.6 2.5 2.1 86
India 4.1 2.7 1.4 51
Brazil 2.9 1.2 1.7 147
All countries 571.1 341.9 225.3 67
1 All direct trades between a customer and a dealer executed either by telephone or on a single-bank
trading system. 2 In billions of US dollars. 3 In per cent.
Source: 2010 Triennial Central Bank Survey. Table 3
… and has lowered
BIS Quarterly Review, December 2010 35
As client flows through electronic platforms have increased, banks are
matching more trades against each other on their books electronically, thereby
capturing the bid-ask spread. While the top dealers report that in April 2007
less than 25% of trades were internalised in this way, by April 2010 they were
matching 80% or more of customer trades internally. These trades settle on the
bank’s books and are not seen by the marketplace, although they are reported
in the Triennial. The market only sees the hedging activity of the remaining
20%, which reportedly takes place in the electronic broking markets via trading
from financial centres, in particular London.12
Banks match client
Increased competition from electronic platforms, combined with improved
trade processing and settlement systems, have lowered transaction costs.
Lower costs, in turn, make more trading strategies profitable, inducing more
speculative activity and encouraging the entry of new participants in global FX
markets. At one end of the scale, macro hedge funds and other leveraged
investors find it more attractive to trade. At the other end, the smaller trade
sizes of retail investors can now be accommodated. These trends are driving
Increase in FX market turnover driven by algorithmic trading
The growth in electronic execution methods in FX markets has enabled
algorithmic trading. Algorithmic trading is an umbrella term that captures any
automated trades where a computer algorithm determines the order
submission strategy.13 For example, FX dealers use algorithms to
automatically hedge risk in their inventories or to clear positions in an efficient
manner. Customers are increasingly using execution management systems
that break up trades and seek the best market liquidity to reduce market
impact. Hedge funds and proprietary trading desks use algorithms to engage in
macro bets, statistical arbitrage or other forms of technical trading. All these
activities are contributing to the increase in FX turnover.
A key turning point for algorithmic trading in FX markets came in 2004
when the electronic broker EBS launched the service “EBS Spot Ai”, where Ai
stands for automated interface. By providing a computer interface to banks,
EBS enabled algorithmic trading in spot FX markets using the real-time prices
quoted on EBS. In 2005, this service was extended to the major customers of
banks, allowing hedge funds and other traders to gain access to inter-dealer
markets – the deepest and most liquid part of the FX market – via their prime
brokerage accounts with the biggest FX dealers.14
It took off in inter-
dealer FX markets
from 2004 ...
12 Lyons (1997) coined the term “hot-potato trading” to describe the repeated passing of
inventory imbalances between dealers. Hot-potato trading is offered as one explanation for
the high inter-dealer turnover in FX markets.
13 Manual traders may use keypads and electronic monitors to follow markets but the decision to
trade is made by a human, with the trade executed either electronically or by telephone.
14 A bank’s prime brokerage customers trade in the bank’s name using the bank’s existing credit
lines with other dealers. Counterparties may not know the identity of the client, only the name
of the prime brokerage bank that is their counterparty on a trade. For more details, see
36 BIS Quarterly Review, December 2010
Algorithmic trading has boosted growth on multi-bank platforms. For
example, on EBS algorithmic spot trading has been rising steadily from 2% in
2004 to 45% in 2010 (Graph 3, centre panel). Algorithmic trading was also
behind the growth of activity in exchange-traded currency futures and options
on the Chicago Mercantile Exchange (CME). The CME first provided an
electronic interface for algorithmic traders in late 2002, leading to sharp
increase in turnover from 2003 onwards. Over the past three years, the CME’s
average daily turnover in FX products has more than doubled to $110 billion
per day (versus $154 billion for EBS) (Graph 3, right-hand panel).
... and has driven
growth on multi-
High-frequency trading (HFT) is one type of algorithmic trading that has
received considerable media attention.15 While HFT emerged over a decade
ago in equity markets, it became an important source of FX growth from 2004.
HFT takes place in the deepest and most liquid parts of the FX market. As the
number of high-frequency traders increased, the traditional profit-making
opportunities from HFT diminished. As a result, the top HFT firms (such as
Getco, Jump Trading and RGM Advisors) have evolved from engaging purely in
price arbitrage on multi-bank ECNs to becoming liquidity providers as well.
trading is one type
of algo trading …
Market estimates suggest HFT accounts for around 25% of spot FX
activity. While many commentators suggest much of the growth in spot turnover
is due to HFT, the contribution of HFT to the increased FX turnover between
2007 and 2010 is not known with precision (Hughes (2010), Lambert (2010)).
Neither the Triennial data on counterparty types nor the data on execution
methods identify HFT. This estimate therefore cannot be verified.
… and accounts for
an estimated 25%
of spot FX
One way to evaluate the importance of HFT to FX market growth is to
identify the instruments, currency pairs and execution methods where this
activity is more likely to show up in the Triennial data. Increased HFT activity
should be associated with: (i) increased trading by the relevant category of
customers; (ii) increased spot turnover due to the ease of electronic trading
and the lowest transaction costs; (iii) increased activity in the main currency
pairs, where turnover is the highest; (iv) increased trading in the United States
and United Kingdom, where high-frequency traders are located; (v) a growth in
trades executed via EBS, Reuters and other multi-bank ECNs; and (vi) HFT
should also be associated with a reduction in average trade size.
Trends in the 2010 Triennial data are consistent with growth in HFT. The
increase in turnover is driven by “Other financial institutions”, the category that
includes HFT. In terms of instruments, most of the increase takes place in spot
trading, which grew by 50% to $1.5 trillion per day in April 2010 (Graph 4, left-
hand panel). The biggest absolute increase over the past three years has taken
place in the US dollar and euro (Table 4). Three quarters of increased spot
trading is located in the United Kingdom and the United States (Table 4). Data
collected by the New York Foreign Exchange Committee also show that
average trade size has declined, consistent with an increase in HFT (Graph 4,
Triennial data are
more HFT …
15 While the 6 May 2010 “flash crash” in US equity markets was initially blamed on HFT, the
report by the US Securities and Exchange Commission relieves HFT of any responsibility,
pointing instead to the order execution algorithm of a US mutual fund.
BIS Quarterly Review, December 2010 37
Evidence of high-frequency trading
Spot FX trading1 Survey by New York Foreign
Exchange Committee Contribution to increase in spot FX
by execution method4
04 05 06 07 08 09 10
Average number of trades (lhs)
Average trade size (rhs)
1995 1998 2001 2004 2007 2010
United States United Kingdom
1 In billions of US dollars. 2 In thousands. 3 In millions of US dollars. 4 In per cent.
Sources: Triennial Central Bank Survey; New York Foreign Exchange Committee. Graph 4
The Triennial data on execution method provide evidence consistent with
HFT as an important source of FX growth. Due to the importance of execution
speed, high-frequency traders need to be located as close as physically
possible to the multi-bank platform’s central matching engine. Given that the
leading multi-bank trading systems such as FXall, Currenex, or Hotspot FX
have their operations in the United States, the increase in activity on these
platforms should rise faster than other execution methods if it is driven by HFT.
Consistent with this hypothesis, one third of the increase in spot trading in the
United States takes place on multi-bank ECNs (Graph 4, right-hand panel). In
the United Kingdom, however, electronic broking systems (such as EBS and
Thomson Reuters Matching) account for a greater share of the increase in spot
trading than multi-bank ECNs. Reuters confirms that the majority of their HFT
Increase in global FX market turnover by currency and location
to FX market
By currency (net-net basis)
US dollar 1,689 266 8 41
Euro 778 162 5 25
Japanese yen 378 91 3 14
All currencies 3,981 657 20 100
By location (net-gross
United Kingdom 1,854 370 9 48
United States 904 159 4 21
Japan 312 62 1 8
All countries 5,056 776 18 100
1 In billions of US dollars. 2 Percentage contribution to the total increase of $657 billion from 2007 to
Source: 2010 Triennial Central Bank Survey. Table 4
… as seen in
38 BIS Quarterly Review, December 2010
clients transact via servers in London to be closer to Reuters’ central matching
Growing importance of retail as an investor class
More than any other customer segment, electronic trading has opened up the
foreign exchange market to retail investors – a trend highlighted already in the
discussion of the 2007 Triennial survey (Galati and Heath (2007)). Trading by
households and small non-bank institutions has grown enormously, with market
participants reporting that it now accounts for an estimated 8–10% of spot FX
turnover globally ($125–150 billion per day). Japanese retail investors are the
most active, with market estimates suggesting this segment represents 30% or
more of spot Japanese yen trading (ie more than $20 billion per day).
Retail investors are
trading up to $150
billion per day …
Retail FX trading takes place over the internet via a new type of financial
institution, the retail aggregator.17 A retail aggregator is a financial firm that
acts as a FX intermediary, aggregating bid-offer quotes from the top FX dealing
banks and facilitating trades by retail investors. Some retail aggregators act
purely as FX brokers, matching retail trades with quotes from banks. Other
retail aggregators combine a broker model with a dealer model; they may act
as the counterparty for some retail trades while passing others directly to the
banks. Based on the quantity of business transacted by their customer base,
retail aggregators secure a commitment from the biggest FX banks to provide
them with tight bid-ask quotes.18 Competing quotes are streamed live to
customers via the retail aggregator’s online platform, typically with a small
markup of one pip or less for the major currency pairs.19 Retail customers
primarily trade spot in the major currencies, although the number of emerging
market currencies offered is growing.
… through online
Retail investors are attracted to FX by the long trading hours, the deep
market liquidity, the low transaction costs and the ability to generate leverage.
Retail customers create leverage by trading via a margin account with the retail
aggregator. The initial cash deposit is used to secure the larger notional value
of their positions, with the margin requirement varying across jurisdictions.
When a trade is executed, the retail aggregator settles it against the margin in
the customer’s account.
… using high
The rapid growth of retail FX trading has led to increased regulation.
Regulators have introduced registration of online FX dealers, raised their
… attracting the
16 With matching engines in all three time zones, high-frequency traders on EBS operate out of a
number of centres, although the United States and United Kingdom are preferred.
17 Examples of retail aggregators are: US-headquartered FXCM, FX Dealer Direct, Gain Capital
and OANDA; European-based Saxo Bank and IG Markets; and Japanese-based
18 As competition has intensified, retail aggregators have begun posting prices out to five
decimal points for the most actively traded currency pairs.
19 For the EURUSD pair, one pip equals 0.0001. On 30 April 2010 the price to buy one EUR
(“the offer”) was $1.3316 and the price to sell (“the bid”) was $1.3315. The bid-ask spread of
one pip is equivalent to $1 on EUR 10,000 ($13,316 – $13,315).
BIS Quarterly Review, December 2010 39
capital requirements and introduced other measures to protect consumers such
as requiring the segmentation of customer funds. The US Commodity Futures
Trading Commission recently reduced the cap on retail leverage from 100:1 to
50:1 for major currencies (and 20:1 for other currencies). Japan’s Financial
Services Authority also reduced leverage to 50:1, with plans to reach 25:1 by
2011. Greater regulation has led to consolidation in this industry, with the
number of retail aggregators in the United States declining from 47 in 2007 to
11 today and in Japan from over 500 in 2005 to around 70 today. In the United
Kingdom and continental Europe, however, there are currently no limits on
leverage and limited regulation, creating the potential for regulatory arbitrage.
Electronic trading is transforming FX markets and encouraging greater trading
by the category of “Other financial institutions”. This broad category includes
smaller banks, mutual funds, money market funds, insurance companies,
pension funds, hedge funds, currency funds and central banks, among others.
Higher trading by other financial institutions is responsible for 85% of the
increase in daily average turnover between 2007 and 2010. Within this
category, the main contribution appears to come from high-frequency traders,
banks trading as clients of the biggest FX dealers and retail investors trading
The investment by all FX participants in electronic execution methods has
increased competition, lowered transaction costs and encouraged the entry of
new participants in global FX markets. These structural changes have also
fuelled the rapid growth of algorithmic trading, particularly high-frequency
trading in spot markets for the major currency pairs. FX instruments where
counterparty credit concerns remain important, such as FX swaps, are proving
more difficult to automate and have grown more slowly. While electronic
execution methods have initially boosted growth in the main financial centres,
this trend is also likely to lift turnover in other countries in the coming years. At
the same time, the relative importance of inter-dealer trading may continue to
decline as banks match more customer trades internally.
40 BIS Quarterly Review, December 2010
Cecchetti, S, J Gyntelberg and M Hollanders (2009): “Central counterparties for
over-the-counter derivatives”, BIS Quarterly Review, September, pp 45–58.
Chaboud, A, B Chiquoine, E Hjalmarsson and C Vega (2009): “Rise of the
machines: algorithmic trading in the foreign exchange market”, The Federal
Reserve Board International Finance Discussion Papers, 2009–980.
Chui, M, D Domanski, P Kugler and J Shek (2010): “The collapse of
international bank finance during the crisis: evidence from syndicated loan
markets”, BIS Quarterly Review, September, pp 39–49.
Galati, G (2002): “Settlement risk in foreign exchange markets and CLS Bank”,
BIS Quarterly Review, December, pp 55–65.
Galati, G and A Heath (2007): “What drives the growth in FX activity?
Interpreting the 2007 triennial survey”, BIS Quarterly Review, December,
Hughes, J (2010): “Innovation drives trading surge”, Financial Times Special
Report – Foreign Exchange, 28 September 2010.
King, M and C Mallo (2010): “A user’s guide to the Triennial Central Bank
Survey of foreign exchange market activity”, BIS Quarterly Review, December,
Lambert, C (2010): “Special report – the $4 trillion market”, Profit & Loss,
October 2010, pp 17–32.
Lindley, R (2008): “Reducing foreign exchange settlement risk”, BIS Quarterly
Review, September, pp 53–65.
Lyons, R (1997): “A simultaneous trade model of the foreign exchange hot
potato”, Journal of International Economics, no 42, pp 275–98.
Melvin, M and M Taylor (2009): “The crisis in the foreign exchange market”,
Journal of International Money and Finance, no 28, pp 1317–30.
Mihaljek, D and F Packer (2010): “Derivatives in emerging markets: what can
we learn from the 2010 Triennial Survey?”, BIS Quarterly Review, December,
BIS Quarterly Review, December 2010 41
42 BIS Quarterly Review, December 2010
Algorithmic trading: Automated transactions where a computer algorithm decides the order-
submission strategy. See also: High-frequency trading.
Bid-ask spread: Difference between the price for buying (ask) and the sell price (bid), which
measures the transaction costs for executing a trade; often used as an indicator of market liquidity.
Broker: A financial intermediary who matches counterparties to a transaction without being a party
to the trade. The broker can operate electronically (electronic broker) or by telephone (voice
Carry trade: A trading strategy where low-yielding currencies are sold to finance the purchase of
Central counterparty (CCP): An independent legal entity that interposes itself between the buyer
and the seller of a security, and requires a margin deposit from both sides.
Counterparty credit risk: The risk that a counterparty will not settle an obligation in full value,
either when due or at any time thereafter.
Dealer (or market-maker): A financial institution whose primary business is entering into
transactions on both sides of markets and seeking profits by taking risks in these markets.
Electronic communication network (ECN): Generic term for a type of computer system that
facilitates electronic trading, typically in over-the-counter markets. Orders are typically entered into
the ECN via the internet or through a private electronic network.
High-frequency trading (HFT): An algorithmic trading strategy that profits from incremental price
movements with frequent, small trades executed in milliseconds for investment horizons of typically
less than one day. See also: Algorithmic trading.
Interdealer market: The market where FX dealers trade with each other, either bilaterally or
through brokers. Also called the “interbank market”, due to the dominance of banks as FX dealers.
Margin account: An account that allows customers to buy securities with money borrowed from a
financial intermediary. The customer’s cash deposit in the account is called the margin.
Market liquidity: A characteristic of the market where transactions have a limited impact on prices
(“price impact”) and can be completed quickly (“immediacy”).
Multi-bank trading system: An electronic trading sytem that aggregates and distributes quotes
from multiple FX dealers.
Prime brokerage: A service offered by banks that allows a client to source funding and market
liquidity from a variety of executing dealers while maintaining a credit relationship, placing collateral
and settling with a single entity.
Reporting dealer: A bank that is active in FX markets, both for its own account and to meet
customer demand, and participates in the Triennial survey.
Retail aggregator: A term used for online broker-dealers who aggregate quotes from the top FX
dealers and provide them to retail customers (individuals and smaller institutions).
Settlement risk: The risk that one of the counterparties to a transaction does not deliver payment.
Single-bank trading system: A proprietary electronic trading system operated by an FX dealer for
the exclusive use of its customers.