Informed trade in spot foreign exchange markets: an empirical investigation

Journal of International Economics 61 (2003) 307–329www.elsevier.com/ locate/econbase

I nformed trade in spot foreign exchange markets: anempirical investigation

*Richard PayneDepartment of Accounting and Finance and Financial Markets Group,

London School of Economics, Houghton Street, London WC2A 2AE, UK

Received 31 August 2002; received in revised form 18 September 2002; accepted 5 December 2002

Abstract

This paper presents new evidence on information asymmetries in inter-dealer FXmarkets. We employ a new USD/DEM data set covering the activities ofmultiple dealersover one trading week. We utilise and extend the VAR structure introduced in Hasbrouck [J.Finance 46(1) (1991) 179] to quantify the permanent effects of trades on quotes and showthat asymmetric information accounts for around 60% of average bid-ask spreads. Further,40% of all permanent price variation is shown to be due to transaction-related information.Finally, we uncover strong time-of-day effects in the information carried by trades that arerelated to the supply of liquidity to D2000-2; at times when liquidity supply is high,individual trades have small permanent effects on quotes but the proportion of permanentquote variation explained by overall trading activity is relatively high. In periods of lowliquidity supply the converse is true—individual trades have large permanent price effectsbut aggregate trading activity contributes little to permanent quote evolution. 2003 Elsevier B.V. All rights reserved.

Keywords: Exchange rates; Market microstructure; Asymmetric information

JEL classification: C22; F31; G15

1 . Introduction

Prior to the 1990s, analysis of the causes of exchange rate movements was a

*Tel.: 144-20-7955-7893; fax:144-20-7242-1006.E-mail address: [email protected](R. Payne).

0022-1996/03/$ – see front matter 2003 Elsevier B.V. All rights reserved.doi:10.1016/S0022-1996(03)00003-5

mailto:[email protected]

308 R. Payne / Journal of International Economics 61 (2003) 307–329

field that was firmly in the hands of macroeconomists. Exchange rate models,based on the goods and asset market approaches were set out and tested using lowfrequency (e.g. monthly or quarterly) data on exchange rates and macroeconomicfundamentals. However, these tests most often revealed that the fundamentals wereless important for exchange rate determination than models predicted. Theexplanatory power of macroeconomic data for exchange rates was poor and theforecasting power of regressions based on fundamentals was less good than that ofa simple random walk. A classic reference along these lines is Meese and Rogoff(1983).

This failure has led, in the last decade or so, to increasing attention being paid tomodels of FX market activity and exchange rate determination based on marketmicrostructure analysis. This large and growing literature places the process bywhich currencies are actually exchanged in centre stage and focusses on the impactof heterogeneities in the trading population for prices and traded quantities.

A key source of heterogeneity in standard microstructure finance is information-al—some agents are assumed to be better informed about future asset prices thanothers (Glosten and Milgrom, 1985). On an empirical level, such informationalasymmetries have several important implications. First, faced with the possibilityof trading with a better-informed individual, uninformed liquidity suppliers widenthe bid-ask spreads that they charge. This allows them to recoup the lossesinflicted upon them by insiders from uninformed individuals. Second, and moreimportantly in the current context, transaction activity carries information and thustrades permanently alter prices. Episodes in which aggressive traders tend to bebuying a given currency will lead to its price rising while the converse is trueduring episodes of aggressive sales. This second prediction is vital—it opens achannel through which transaction activity in FX markets might play a role inexchange rate determination, a feature that is entirely absent from standardmacroeconomic exchange rate models. The current study seeks to assess theimportance of this channel.

Recent empirical papers that also focus on the explanatory power of currencytrading activity for exchange rate changes include Lyons (1995) and Yao (1998).Both of these studies use data from single FX dealers to demonstrate that spreadscontain an asymmetric information component. Lyons (1996) extends his priorwork by examining the role oftime in the relationship between trades and quotes.He finds that trades occurring in periods when the market is active convey lessinformation than those consummated when the market is quiet. This is interpretedas consistent with his ‘hot potato hypothesis’ by which high interdealer volumesare generated more by inventory rebalancing than exploitation of private in-formation. Most recently, studies by Evans and Lyons (2001) and Evans (2001)provide strong evidence for an information content to inter-dealer FX order flowusing 4 months of data on direct (i.e. non-brokered) FX trading activity.

Theoretical models that focus on the information contained in inter-dealer spotFX trading activity can be found in Lyons (1995) and Perraudin and Vitale (1996).

R. Payne / Journal of International Economics 61 (2003) 307–329 309

In these models dealers receive private signals of future exchange rate evolution1,2from their customer (i.e. non-dealer) order flow. On an institutional level, this is

possible as customer–dealer trade is entirely opaque (i.e. dealer B cannot observethe customer order flow arriving at dealer A and vice versa). The customer orderflow arriving at dealers may be informative for a number of reasons. First, andmost blatantly, a given dealer might have an intervening central bank as acustomer and thus may learn about future interest rates from the central bank’sorder. However, such occurrences are likely to be rare and hence this channel is

3not ideal for arguing that customer orders contain information in general. Asecond rationale for the information content of customer order flow comes fromarguing that information regarding future exchange rate fundamentals, for exampletrade balances, is dispersed among individual customers. A given dealer observesthe trading behaviour of a group of individual customers and, thus, from theiraggregate trading activity receives a signal regarding future fundamentals whichcan be exploited in inter-dealer trade.

Note that in both of the prior examples, customer order flow is informativeabout exchange rate fundamentals—i.e. interest rates or trade balances. Anotherclass of models, see Evans and Lyons (2001), generates customer trading activitythat forecasts future prices because it is informative about risk premia. These‘portfolio shifts’ or portfolio balance models generate permanent price shifts dueto risk-aversion. Assume that the non-dealer segment of the market experiences aportfolio shift that requires currency trade. Further assume that there is noexpectation that this trade will be reversed such that the aggregate inventoryimbalance (and risk) foisted upon the market is permanent and undiversifiable.Assuming risk averse agents operating in the FX markets then delivers apermanent change in prices due to a permanent change in risk premia. Dealers areinformed in this setting as they see a signal of the aggregate portfolio shift through

1Throughout this paper we will repeatedly refer to trades, volumes and order flow. Volume has itsusual definition as the sum of all Dollar quantities traded in a given interval. However, each individualtrade may be given a direction according to whether the aggressor (the agent demanding liquidity) is abuyer or a seller. Order flow is the difference between buyer-initiated and seller-initiated tradingactivity. It can be measured in Dollar terms (i.e. as the volume of buyer-initiated activity less thevolume of seller-initiated) or simply in terms of the number of trades (i.e. as the number of buyer- lessseller-initiated trades).

2Another interesting recent paper by Chakrabarti (2000) explores the possibility that dealers alsolearn from the quotes of other dealers. Whilst this is interesting, it is not possible to empiricallyevaluate this proposition with our data and thus we leave it to one side.

3Peiers (1997) examines USD/DEM dealer quoting around Bundesbank intervention events, findingthat Deutsche Bank is a price leader at these times. Given that Bundesbank intervention operations arecarried out via Deutsche Bank, these results lend credence to the notion that observing central banktrades conveys price-relevant information. Corroborating evidence is provided by Naranjo andNimalendran (2000), who demonstrate that the adverse selection component of FX spreads increasesaround central bank intervention events.


their customer order flow, before the size of the shift becomes public. Again,dealers can exploit their signals in trade in the inter-dealer market.

As stated above, for the current paper the key empirical implication of thepreceding models is that trading activity in the inter-dealer market will have apermanent affect on interdealer quotes. This permanent effect is generated by theinformation carried by inter-dealer trades (regardless of whether this information isabout future fundamentals or future risk premia). We test for this permanent priceeffect using the bivariate VAR model for quote revisions and signed tradesintroduced in Hasbrouck (1991a). Further, whilst earlier studies have demonstratedthat at least some FX trades carry information, none has computed the aggregateimpact of such information. Use of the variance decomposition presented inHasbrouck (1991b) allows us to calculate the proportion of all informationentering the quotation process via order flow and hence address this issue. Finally,we examine variations in the information content of trades with the pace of themarket. Theoretical contributions such as Admati and Pfleiderer (1988) and Fosterand Viswanathan (1993) predict correlations in the intra-day variation of transac-tions costs, volume and the intensity of informed trade which we evaluate via atime-of-day subsample analysis of the D2000-2 data. We refine the time-of-dayanalysis by modifying the basic VAR structure to allow for dependence of theparameters on D2000-2 liquidity measures.

One of the main innovations of this work is the use of a new data set oninter-dealer USD/DEM trades, drawn from an electronic brokerage called D2000-2. Whilst earlier work in this area has employed data based on the operations ofsingle dealers, the D2000-2 data reflect the interactions of multiple traders. Assuch, these data provide broader coverage of inter-dealer activity. Further, theD2000-2 data can be used to construct proxies for the liquidity of the FX marketas a whole. D2000-2 operates as a closed, electronic order book and every limitand market order entered onto the system are available from the data. Liquidityand depth measures can be constructed from the limit order data and used asconditioning variables in the analysis of the effects of private information.

Our main results are as follows. First, our estimates imply that over 60% of theD2000-2 spread can be thought of as compensation for informed trade. Anunexpected market buy, for example, leads to an upwards equilibrium quoterevision of 1 pip (i.e. DM 0.0001) on average. The 60% figure derives fromcomparing this 1 pip permanent price impact to one half of the average bid-askspread on D2000-2. Further, we estimate that around 40% of all informationentering the quotation process does so through order flow, a figure which iscomparable in magnitude to equivalent measures from equity market studies.

The 40% number quoted above is important in that it quantifies the relativeimpact of trade-related versus non-trade-related information on exchange rateevolution. However, given that the data we study here cover only one of severalvenues for inter-dealer FX trade we should clarify its interpretation. Our resultdoes not imply that 40% of the information relevant to long-run USD/DEM


determination is revealed on D2000-2. The number is derived from the correlationbetween D2000-2 order flow and permanent USD/DEM movements. If we were toassume, quite reasonably, that all inter-dealer FX venues shared the samepermanent price component, and further that all order flows were perfectlycorrelated, then we would get the same value for the size of the trade correlatedcomponent from each venue. Thus in this case 40% would be a market-widefigure. We would suggest that, in the current context, the size of the tradecorrelated component should be interpreted along similar lines. However, there isthe possibility that informed FX trade is more prevalent on other (less transparent)trading venues. In this case, our 40% number might be interpreted as a lowerbound on the impact of private information in FX markets.

Finally, impact of informed trade on D2000-2 quotes is shown to dependstrongly on the level of market activity. When the D2000-2 order book is relativelythin and volume low (i.e. from the late GMT afternoon to the early GMTmorning,) an unexpected trade has a much larger permanent effect on quotes thanin peak trading periods. This result is consistent with the empirical analysis ofLyons (1996) and the theoretical predictions of Admati and Pfleiderer (1988). Amore direct examination of the relationship between informed trade and marketconditions is obtained from some extended VAR results. These demonstrate thatthe permanent impact of informed trade on quotes and the supply of limit orders toD2000-2 are negatively and non-linearly related. However, it should be noted thatwhile we show that individual trades have larger permanent price effects in periodsof lower trading activity or liquidity supply, the variance decompositions tell usthat the aggregate contribution of transaction activity to permanent exchange ratedetermination is low at these times. Thus these periods should not be consideredthose in whichprice discovery is particularly intensive.

The rest of the paper is set out as follows. Section 2 introduces the basicfeatures of the D2000-2 data set and Section 3 details the empirical methodologyemployed in the current study. Section 4 presents the empirical findings from theVAR estimations. Finally, Section 5 concludes and presents ideas for further work.

2 . The dealing system and the data

Around 80% of all spot FX trade is inter-dealer, with the remainder consistingof trade between dealers and non-financial customers or dealers and non-dealerfinancial customers. Until fairly recently, all inter-dealer trade, both direct andbrokered, was carried out over the telephone. Dealers would call one another torequest price information and consummate trades or, alternatively, might callhuman brokers, also known as voice brokers, to express their trading interests. Oneimplication of this was that, aside from the triennial BIS surveys of FX marketactivity, no consolidated source of FX trade information existed. Further, the orderflow information available to dealers themselves was limited. Indicative quote


information was available via a number of screen based systems and reports ofvoice brokered trades were broadcast over intercom systems, but there were noindications of market wide order flow.

In the early 1990s, however, a shift away from telephone based trade occurredwith the introduction of several electronic trading systems. Reuters opened anetwork called D2000-1, facilitating direct, bilateral inter-dealer trade. Evans(2001) and Evans and Lyons (2001) analyse transactions data from this network.A further two electronic broking systems came into being—that run by the EBSconsortium and the Reuters D2000-2 system. These systems have grown quickly,driving the voice brokered portion of trade down considerably. In June 1997 EBSclaimed to handle 37% of all brokered trade in London, with the D2000-2 market

4share commonly assumed to be similar. Based on this figure and on BIS surveyinformation, one might estimate that the share of spot inter-dealer USD/DEMvolume passing through D2000-2 in 1997 was around 10%. The data used in thisstudy cover all USD/DEM trade on D2000-2 over the week 6th–10th October1997. Around 30,000 transactions occurred during this time with total volumeapproaching $60bn.

2 .1. The D2000-2 dealing system

D2000-2 is an electronic order driven system. Liquidity is supplied to thesystem via limit buy and sell orders and is drained from the system in two ways;first, through market buy and sell orders and direct crossing of limit orders and,

5second, through voluntary cancellation of limit orders. Transaction consummationis governed by rules of price and time priority subject to one proviso. Participantsin the system must bilaterally negotiate credit lines if they wish to trade. Thustrades sometimes execute outside the inside spread due to lack of credit linesbetween limit and market order submitters.

A subscriber to D2000-2 sees the following items on the trading display. Firstthere is an indication of the most competitive limit buy and sell prices in thesystem along with the quantities available at those prices. No information onsubsidiary limit orders (i.e. buy/sell orders with prices below/above the extantbest price) is displayed. The screen also indicates the last transactions whichoccurred in a given currency pair, detailing price and volume. The aboveinformation is simultaneously available for multiple exchange rates. From our

4At the current time, that is mid-2002, the market shares of D2000-2 and EBS are certainly notsimilar. Reuters has the greater market share in Sterling-related exchange rates but EBS appears to haveachieved market dominance in most others, including the key Euro–Dollar rate.

5By ‘direct crossing of limit orders’ we mean the situation in which the D2000-2 order book containsa limit buy with price greater than or equal to that of a limit sell. In this case the system automaticallytransacts the overlapping quantity. Thus, we can treat the entry of limit orders that result in crosses assimilar to market order entries.


USD/DEM data set, one can reconstruct all of the information available tosubscribers. Additionally, the data set contains information on every limit orderentered on D2000-2.

2 .2. The data set

The raw D2000-2 data feed consists of just over 130,000 data lines. Each linecontains 10 fields detailing the type of event to which it refers, timestamps with aone hundredth of a second accuracy, price and quantities. Just over 100,000 ofthese data lines are limit order entries (with timestamps for entry and exit times, abuy/sell indicator, quantity available, quantity traded and price.) The rest of thedata consists of market order entries. The market order lines give the quantitytransacted and price, a timestamp and whether buyer or seller initiated. From theraw data we construct an event time data set including the following variables; themidquote (i.e. the average of best limit buy and limit sell prices); a signedtransaction indicator variable; signed transaction quantity (in $m); the insidespread; aggregate buy and sell order book size and the number of buy and sell

6limit orders outstanding. The first panel of Table 1 presents descriptive statisticsfor the market activity variables for seven non-overlapping subsamples of alltrading days. The first six of these subsamples consist of observations from 2 hoursegments of each day, covering the period from 6 to 18 GMT. The seventhsubsample represents data from GMT overnight periods, 18 to 6 GMT.

The main feature of these statistics is that all series have strong repetitiveintra-day patterns. The number of limit orders outstanding and aggregate size onthe book broadly follow an inverted U-shape across the GMT trading day.Transaction frequency and volume data show a similar pattern, aside from a lull inactivity in the period from 10 to 12 GMT. D2000-2 liquidity, as measured by thepercentage spread, follows the inverse pattern to transaction activity. It isparticularly noticeable that D2000-2 is extremely illiquid between the hours of 18and 6 GMT. Spreads are very high, order book size limited and in our data thisinterval accounts for less than 5% of transaction activity. During the complemen-tary portion of the trading day the spread is very small (with a modal value of onepip) and the D2000-2 order book is very deep.

Panel (b) of Table 1 contains summary statistics for the percentage change inthe midquote for our seven subsamples. Again, the effects of the intra-day in

6We define an event as a revision in the best limit buy or sell price or the occurrence of a transaction.The transaction indicator is signed positive when the aggressor (i.e. the market order trader) is buyingand negative when the aggressor is selling. The transaction indicator is zero when there is a revision ineither of the best limit prices without an accompanying trade. Transaction indicators were constructedboth including and excluding crosses. When included, crosses are signed by treating the latest enteringlimit order as the aggressor. Limit buy (sell) order book size is defined as the aggregate quantity, in $m,outstanding across all buy (sell) limit orders.

314R

.P

ayne/

Journalof

InternationalE

conomics

61 (2003) 307–329

Table 1Summary statistics for D2000-2 order book and return data

Panel (a): Order book statistics¯Sample Obs. Bids Offers Q Q s Deal Vol. Sizeb o

6 to 8 GMT 10816 31.14 31.25 55.24 58.68 0.015 4921 8742 1.788 to 10 GMT 11587 46.63 47.02 91.39 89.82 0.011 5751 11036 1.92

10 to 12 GMT 10446 47.74 42.15 99.84 79.29 0.017 4654 8600 1.8512 to 14 GMT 17339 50.22 46.77 95.31 80.79 0.014 7519 13888 1.8514 to 16 GMT 12088 41.65 33.31 86.59 57.37 0.018 4101 7302 1.7816 to 18 GMT 3333 20.04 14.46 54.89 22.04 0.042 671 1080 1.6118 to 6 GMT 4938 9.33 7.43 25.45 15.06 0.076 766 1155 1.51

Panel (b): Return statistics2Sample Mean Var. Skew Kurtosis r Q(5) Q (5)1

6 to 8 GMT 0.00002 0.00006 0.40 92.5 20.38 1619.39 1224.928 to 10 GMT 20.00000 0.00003 0.10 34.2 20.35 1459.78 1254.48

10 to 12 GMT 20.00009 0.00013 20.38 127.2 20.39 1632.87 1131.3312 to 14 GMT 0.00001 0.00005 0.04 55.9 20.34 2086.08 1910.4414 to 16 GMT 0.00004 0.00007 20.05 180.9 20.35 1492.81 1066.2016 to 18 GMT 20.00013 0.00036 0.11 18.9 20.42 601.19 1646.2718 to 6 GMT 0.00001 0.00230 0.05 34.5 20.50 1267.84 1814.57

Note: panel (a) of the table gives basic statistics for order book data. Columns headedBids andOffers give the average number of orders outstanding across a given¯sample. The next two columns give average aggregate quantity outstanding on both sides of the order book in $m.s is the average percentage spread in a given

subsample. The final three columns give the total number of transactions, total volume traded in $m and mean transaction size in $m in each subsample. Panel (b)gives summary statistics for midquote returns for the time-of-day-based subsamples. The first four columns of the table give the first four sample moments of thereturn series. The fifth column gives the first order return autocorrelation. Column six presents fifth order Box–Ljung statistics for returns and the final column givesfifth order Box–Ljung statistics for squared residual returns, where residual returns are created by filtering an MA(1) from actual returns. For the tests reported in thefinal two columns, the asymptotic 5% and 1% critical values are 11.07 and 15.09 respectively.


D2000-2 activity are apparent. Both return dependence and volatility areinverselyrelated to traded volume. An information flow model of trading would suggest theopposite relationship. Our explanation for the observed correlation is that it is dueto D2000-2 liquidity variation: while in peak trading periods more information isimpounded into midquotes and midquotes change more frequently, in less activeperiods the thinness of the D2000-2 order book causes measured return volatilityto be high around trades or limit order cancellations.

The figures in Table 1 imply that D2000-2 only operates effectively duringEuropean and North American trading hours. Hence, in the empirical analysispresented in Section 4, observations from the 18 to 6 GMT period are omitted.

3 . Empirical methodology

As discussed in the previous section, D2000-2 is a multi-lateral order drivensystem. This implies that the empirical models used in Lyons (1995) and Yao(1998) are inapplicable here as they are based on an underlying quote-driven,single-dealer structure. Instead, we employ the reduced form VAR in trades andquote revisions developed in Hasbrouck (1991a) and Hasbrouck (1991b). Thisframework is not predicated on any particular underlying microstructure modeland has been used in the analysis of entirely order driven markets in de Jong et al.(1995) and Hamao and Hasbrouck (1995). Information-based trade is identifiedvia a positive and significant long-run response of quotations to transactionactivity, in line with the theoretical argument presented in Section 1. Further, thevariance decomposition presented in Hasbrouck (1991b) allows one to evaluate theamount of information entering the FX quotation process that is trade-related andhence the contribution of information-based trade to price discovery.

There are two main assumptions which underlie the application of thisframework to the current data set. The first of these is that informed agents exploittheir advantage through the use of market orders rather than limit orders.Non-informed agents, on the other hand, submit either market or limit orders toD2000-2 depending on their desire for execution speed. This implies that privateinformation canonly influence prices through unexpected trading activity. Theassumption may be justified by noting that an informed agent submitting a limitorder is noisily advertising his beliefs and hence possibly eroding his advantage.Moreover, using limit orders to exploit (short-lived) information advantages isrisky due to the possibility of non-execution. Finally, the analysis of informedorder placement strategy presented in Harris (1998) suggests that informed agentsshould prefer to trade via market order in fast paced, liquid markets. The spot FXmarkets certainly exhibit these features lending credence to our assumption.

The fact that everyone trading on D2000-2 is a dealer makes the distinctionbetween informed and uninformed agents less clear than in other examples. In theIntroduction we argued that customer order flows were the source of information


asymmetries between dealers. Thus one way to discriminate between informed andnon-informed dealers is to appeal to variation in the size and quality and size of

7dealer customer bases. It should also be borne in mind that there is nothing herethat requires the identity of informed traders to be the same over time. At certainpoints in time, dealers with unusually good customer order flows may act asinformed agents in the inter-dealer market (and, via the preceding arguments, tradeby market order) whilst at other times those same dealers are essentiallyuninformed (and supply liquidity to D2000-2).

The second required assumption is that public information is immediatelyreflected in quotes. If this was not the case, traders observing information releasescould form profitable trading strategies which would generate correlation betweentrades and subsequent quote revisions in the absence of private information.However, given the levels of liquidity and competition in FX markets thisassumption would seem to be reasonable.

A final point that should be discussed here is the attractiveness of D2000-2 as avenue for informed trade. As D2000-2 competes for order flow with several othertrading venues we should ask where informed dealers are most likely to trade.Given the microstructural similarities between D2000-2 and EBS, this discussionshould focus on the implications of differences between electronically brokeredtrading and direct trading (whether electronic or over the telephone). The formeroffers trading opportunities that are pre-trade anonymous but which are broadcastto the rest of the market. The latter offers trading opportunities that are non-anonymous but which remain private to the two counterparties. In our eyes, then,both avenues for trade have an advantage—the pre-trade anonymity of electronicbrokers versus the post-trade opacity of direct trading. Thus, on a theoretical level,the venue of choice for informed trade is not clear cut and we would argue thatthere is no reason to believe that informed traders would always avoid trading onD2000-2.

3 .1. The VAR model

Denote the percentage change in the midquote byr and letx represent a vectort t

of transaction characteristics, wheret is an event–time observation counter. Thebasic VAR formulation used in our empirical work is as follows:

P P

r 5O a r 1O b x 1´ (1)t i t2i i t2i 1ti51 i50

7Citibank, as perhaps the biggest player in customer–dealer FX trade, might be thought of as aninformed participant in the inter-dealer market whilst a smaller bank with a much smaller customerbase might be considered uninformed. Along the same lines, an interesting recent paper that linkseffects of a dealer’s trades on the market to that dealer’s reputation is Massa and Simonov (2001).


P P

x 5O g r 1O d x 1´ (2)t i t2i i t2i 2ti51 i51

where the errors are to be zero mean and mutually and serially uncorrelated at all2 2 2leads and lags. Further, we define Var(´ )5E(´ )5s and Var(́ )5E(´ )51t 1t 2t 2t

V.Eqs. (1) and (2) form a general model of the dynamics of trades and quotes and

the interactions between these variables. Note that the VAR is not entirely standardas the contemporaneous realisation ofx enters the return equation. Hence tradest

logically precede quote revisions. This ensures that the innovations to the twoequations are uncorrelated and identifies the VAR.

Given the two assumptions detailed above, the innovations in the VAR can beinterpreted as follows. The innovation to the return equation reflects transitoryquote variation and the effects of public information on quotes. The innovation tothe trade equation represents unpredictable transaction activity and hence thepossibility of information-based trade. Using this classification, the effects ofprivate information on quotations are easily retrieved. First, we invert the VAR toretrieve the VMA representation:

r a(L) b(L) ´t 1t5 (3)S D S DS Dx c(L) d(L) ´t 2t

2 kwhere, for example,a(L)5 a 1 a L 1 a L ? ? ? 1 a L . Given the lack of0 1 2 k

correlation between the innovations, the coefficients in the VMA lag polynomialsare precisely the impulse responses implied by the VAR. The coefficienta , fork

example, is the effect of a unit return shock on the midquote return at ak periodhorizon. The effects of private information are revealed in theb(L) polynomial.The possibility of informed trade implies that quotes respond permanently to trade

`innovations and henceb(1) (i.e. o b ) should be positive and significant.i50 i

Estimation of the VAR and calculation of the implied impulse response functionswill hence allow us to evaluate the existence of information-related trade onD2000-2. In the empirical analysis reported in Section 4, the order of the VAR waschosen via application of the Schwarz information criterion. The VAR equationswere estimated by OLS and are reported with heteroskedasticity robust standarderrors. The VMA representation was calculated by simulation.

3 .2. The variance decomposition

While the VAR model allows us to quantify the information content of a singletrade, it does not permit one to assess the overall importance of informed trade indetermining the evolution of the exchange rate. Hasbrouck (1991b) presents avariance decomposition for returns, based on the VAR structure above, whichpermits retrieval of the variance of the permanent component of midquotes, plus


the proportion of permanent variation related to order flow. The overall influenceof informed trade can be estimated with the latter measure. Denoting the (log) ofthe midquote asq , the following structure is assumed:t

q 5m 1 s (4)t t t

Eq. (4) decomposes the midquote into a random walk component (m ) and at

mean-zero stationary term (s ). Hence,t

2m 1m 1v , v |N(0,s ), E(v v )5 0 for t ± st t21 t t w t s

and lim E s 5 lim Es 50. From an economic perspective,m can bek→` t t1k k→` t1k t

thought of as the fundamental or full-information price process. The transitorycomponent,s , represents that portion of the current midquote generated by anyt

non-information based microstructure effect, e.g. price discreteness, digestioneffects or inventory control.

2The key parameter in the preceding formulation iss which measures variationw

in the permanent component. This can be estimated by equating the returnrepresentation implied by Eq. (4) (using the fact thatr 5Dq ) and the returnt t

equation from the VMA. Furthermore, variation in the permanent component dueto order flow alone can also be retrieved. These measures are calculated as:

` ` ` 22 29s 5 O b V O b 1 11O a s (5)S D S D S Dw i i i

i50 i50 i51

` `

9s 5 O b V O b (6)S D S Dwx i ii50 i50

Standard errors for these two variance estimates can be computed via a residualbased bootstrap of the estimated VAR system. An economic interpretation of Eqs.(5) and (6) is as follows. Public information events are incorporated into theexchange rate via the return innovation,´ . The permanent effect on midquotes of1t

`a unit return shock is given by unity (the contemporaneous impact) pluso ai51 i

and hence the variation in the permanent component implied by public informationevents is given by the second term on the right hand side of Eq. (5). Privateinformation is impounded into the exchange rate via trade innovations with the

`permanent impact of an unexpected unit trade given byo b in the case wherei50 i

x is scalar. For vectorx , the variation in the permanent component driven byt t

trade innovations is thus the first term on the right hand side of (5).

3 .3. Non-linear effects in the trade–quote relationship

While the VAR structure presented in Section 3.1 provides a fairly robustcharacterisation of the determination of trading activity and quote revisions, itrestricts the relationship between these two variables to be invariant to underlying


market activity. Theoretical considerations, however, suggest that the pace of themarket will impact upon the response of quotes to trades. Admati and Pfleiderer(1988) suggest that there should be a negative correlation between the informationcontent of trades and overall volume, for example, and the empirical results ofLyons (1996) might be considered consistent with this.

We allow for the possibility that the dynamic relationship between quotes andtrades varies with market conditions by using anarranged VAR estimation.Specifically, consider a weakly exogenous candidate measure of the pace of themarket,z , and denote byD the matrix of time-series data onz , returns, tradest t t

and their respective lags i.e.D 5 hz , r , r , . . . , r , x , x , . . . , x j. Now,t t t t21 t2P t t21 t2P

consider a sort ofD according to an increasing order sort ofz . Denote the sortedt t

data set byD . One way to investigate variation in the VAR parameters withs

market activity would be to estimate Eqs. (1) and (2) using successively largernumbers of sorted observations via recursive least squares. Evidence of parameterchange could then be evaluated via plots of the recursive coefficient estimates orthrough formal tests for parameter instability.

We follow a similar route to the above but we run a series of regressions wherethe number of observations in each regression is fixed and this fixed window ismoved through the sorted data sample. We use a window size of 2000 observationssuch that our first regression uses sorted observations labelleds51–2000 (i.e. thefirst 2000 rows of the sorted data setD ), our second regression uses rowss

101–2100 ofD , the next regression uses rows 201–2200 etc. This techniques

provides similar information to that outlined in the previous paragraph but theseries of estimated parameters is not ‘smoothed’ by the use of an increasingsample size. Plots of the parameter estimates are used to gain insight into theexistence of non-linearities in the VAR structure. The first lag of the number andaggregate size of outstanding limit orders are employed asz in our analysis.t

4 . Results

This section presents the results from estimating the VAR models in Section 3.1and Section 3.3 and the variance decomposition in Section 3.2. We begin withresults for the entire D2000-2 trading day data set. Results for time-of-day basedsubsamples are discussed next. Finally we analyse the effects of market activity onthe information content of trades via the arranged VAR models. The basicvariables included in our VARs were percentage midquote returns and a signedtransaction indicator (which takes the values21, 0 and11). In the constructionof the transaction indicator only market orders were used, limit order crosses wereignored. Inclusion of crosses in the definition of the transaction indicator changesthe results only marginally. A signed volume variable was similarly constructed.

The first few columns in the top row of row of Table 2 give a summary of therelevant VAR parameters estimated using all trading day observations. Some

320R

.P

ayne/

Journalof

InternationalE

conomics

61 (2003) 307–329

Table 2Summary of VAR results for trading day and subsamples

Subsample VAR results Variance decompositions2 2 2 2 2Lag SIC o b x (b ) QIR Q(10) s /s s /s s /si i i wx v v r wx r

6 to 18 GMT 8 89877 0.00668 4178.7 0.00518 36.1 0.41 (0.015) 0.27 (0.009) 0.11

6 to 8 GMT 5 28425 0.00565 837.8 0.00502 42.4 0.43 (0.030) 0.32 (0.021) 0.148 to 10 GMT 5 216964 0.00406 1060.1 0.00404 19.9 0.47 (0.027) 0.42 (0.022) 0.20

10 to 12 GMT 5 2106 0.00758 561.5 0.00583 70.0 0.32 (0.034) 0.25 (0.019) 0.0812 to 14 GMT 4 27213 0.00548 1344.1 0.00490 55.6 0.36 (0.020) 0.39 (0.018) 0.1414 to 16 GMT 9 24220 0.00750 804.6 0.00564 11.2 0.41 (0.041) 0.23 (0.021) 0.1016 to 18 GMT 5 2406 0.01853 247.1 0.01192 56.7 0.35 (0.046) 0.17 (0.014) 0.06

Note: the table summarises the VAR results from the entire trading day and for the 6 time-of-day subsamples. The column headed ‘Lag’ reports the numberof lagsin the VAR, chosen using the Schwarz information criterion, which is reported in the second column.o b is the sum of the asymmetric information coefficients fromi i

the VAR. The following column gives a Wald test statistic for the null that all are zero. The asymptotic 5% and 1% critical values for this test are 3.84 and 6.63respectively. QIR denotes the long-run quote impulse response implied by the VAR and the column headedQ(10) gives 10th order Box–Ljung statistics for the VAR

2residuals. The asymptotic 5% and 1% critical values for this test are 18.31 and 23.21 respectively.s /s is the proportion of the permanent midquote variance that iswx v2 2 2trade-correlated.s /s is the size of the permanent component as a proportion of total return variance.s /s is the ratio of the trade correlated component to overallv r wx r

return variation. These figures are calculated using the expressions in Eqs. (5) and (6). Numbers in parentheses in the first two columns are bootstrapped standarderrors for the ratios. They are calculated from a standard, residual-based bootstrap of the model using 500 bootstrap replications.


general comments on those parameters not presented are as follows. First, quotereturns demonstrate significant negative autocorrelation at all included lags. Thetransaction indicator displays strong positive autocorrelation. This indicates runs inbuying and selling activity and may be due to dealers splitting large market orders

8or possibly order flow imbalances induced by informed trade. Finally, the effectsof lagged quote returns on transaction direction are mixed. Although marginallysignificant as a group, these coefficients were generally individually insignificant.

2 2The R in the return equation was 0.22 whilst the trade equationR wasapproximately 0.075.

From the current perspective, the key parameters in the VAR formulation arethose labelledb in Eq. (1) i.e. the effects of trades on current and subsequenti

midquote returns. As shown in row one of Table 2 the sum of these coefficients ispositive. Moreover, each individual coefficient is positive and all are statisticallysignificant. Hence, a market buy tends to increase quotes. Computing the VMArepresentation and calculating the equilibrium midquote impulse response showsthat an unexpected market buy leads to an upward quote revision of around

90.005% on average. The average percentage bid-ask spread on D2000-2 is around0.016%. Expressing the equilibrium price impact of a trade as a fraction ofone-half of the bid-ask spread gives an estimate of the contribution of asymmetricinformation to spreads. In this case, the numbers tell us that around 60% of thespread is compensation for asymmetric information.

This 60% figure is very high relative to recent estimates of similar quantitiesfrom equities markets. For example, Huang and Stoll (1997) report averageasymmetric information components between 10 and 20% for the 20 stocks of theUS Major Market Index. At least some of the discrepancy might be explained byfundamental microstructural differences between the NYSE and the electronicallybrokered segment of the inter-dealer FX market. One would conjecture that bothorder processing costs and inventory control costs would be larger for anintermediated equity market, such as the NYSE, where trade frequency is lowrelative to that in FX, for example. The fact that spreads in the Huang and Stoll(1997) data are around 25 basis points on average, while in our inter-dealer FXdata the average percentage spread is less than 1 basis point, lends credence to thisconjecture. Nonetheless, the fact that the asymmetric information component is solarge in our data is noteworthy.

It should be noted that we also experimented with the use of trade size variablesin the trade description vector (x ). One such experiment involved the use oft

signed volume and signed squared volume in addition to trade direction. In a

8In many other data sets a similar result would be found due to the data recording one transaction inwhich a market order executes against a number of limit orders as a series of trades, one for each limitorder affected. Note that in the D2000-2 data this is not the case: one market order appears as one tradein the data set, regardless of how many limit orders it executes against.

9This translates to a 1 pip (i.e. DM 0.0001) quote increase approximately.


second set of experiments we examined whether the largestk% of trades (wherewe variedk from 5 to 25) had return impacts that differed from those of all othertrades. In contrast to the results of Yao (1998), the size variables were generallyinsignificant. This is likely to be due to the fact that there is very little variation intrade size on D2000-2. Over 90% of D2000-2 trades are for less than $5m suchthat the data may not contain the power needed to identify a size effect.

The result presented in the preceding paragraphs is key—market orderspermanently alter D2000-2 quotes and we take this as evidence that they areinformation-motivated. Of course, given that our study is based on 5 days of data,it might be argued that the effects we uncover are not permanent in that they dieout over a horizon longer than 1 week. This is a possibility. However, a number oftheoretical and empirical arguments support our viewpoint. First, on a theoreticallevel, effects of order flow on exchange rates that mean-revert over weeks ormonths might indicate profit opportunities, violating market efficiency. Second,corroborating empirical evidence on the long-run/permanent effect of marketorder activity on quotes can be drawn from regressions of returns on order flowsaggregated over fixed calendar time intervals as in Evans and Lyons (2001). Wehave run a set of such regressions having midquote returns for samplingfrequencies ranging from 20 seconds to 1 hour on the left-hand side and aggregateorder flow (i.e. the excess of the number /volume of market buys over market sellsin an interval) on the right-hand side. We choose 1 hour as the lowest samplingfrequency due to the fact that our entire span of data only covers 5 days. Resultsdemonstrate that across all sampling frequencies, the effect of order flow on

2returns is highly significant. Moreover, theR of the regression increases as thesampling frequency decreases such that at a 1 second level, order flow explainsapproximately 70% of all return variation. Thus, we believe that the resultspresented here, coupled with those in Evans and Lyons (2001), provide clearevidence for the information content of order flow and its longer run effects on theUSD/DEM.

The results above confirm the findings of previous research in that a portion oftrade on D2000-2 can be characterised as information-related. A question whichhas not been directly addressed in the literature, however, is the extent to whichsuch asymmetries alter over the trading day. Microstructure theory relevant tointra-day variations in the intensity of informed trade and liquidity can be found inAdmati and Pfleiderer (1988) and Foster and Viswanathan (1990). The endogen-ous information acquisition model of Admati and Pfleiderer (1988), for example,predicts that high volume periods should be characterised by relatively small priceimpacts from trade. This is due to the clustering of discretionary liquidity trade andincreased competition between informed traders in equilibrium. The model ofFoster and Viswanathan (1990) has a similar setup except that the informationadvantage of insiders is assumed to decline over time.

We examine these issues by estimating the VAR separately for the six previouslydefined non-overlapping subsamples of the trading day. The results, are given in


rows 2 to 7 of Table 2, correspond with the predictions made by Admati andPfleiderer (1988). First note that Table 1 shows that the segments of the tradingday with greatest volume are those with lowest percentage spreads. Admati andPfleiderer (1988) construct a batch trading model and, hence, it does not contain aspread. However, given that their framework predicts more liquid markets in highvolume intervals this result is consistent with their analysis. Further, Table 2indicates that the information content of trading activity follows an intra-daypattern which is the inverse of that followed by volume. Thus, in high volume/liquidity periods, the price response to a trade is relatively low. This is consistent

Fig. 1. The response of the midquote to a unit trade innovation. Notes: these plots show the response ofthe D2000-2 midquote (i.e. the average of the best limit buy and sell prices) to a unit trade innovation.The impulse response functions are calculated via simulation from the VAR estimates summarised inTable 2. Panel (a) presents the impulse response function estimated from all trading day data (i.e. alldata falling in the 6–18 GMT interval. Panel (b) shows the function calculated using data from the8–10 GMT subsample only and, similarly, panel (c) shows the impulse responses implied by the VARestimated using data from the 16–18 GMT interval. The solid line in each panel gives the actualimpulse response function and the dotted lines are a 95% confidence band, estimated using a residualbased bootstrap of the VAR model with 500 bootstrap replications. Thex-axis values give the numberof observations since the trade shock was first felt.


with the ‘hot-potato’ hypothesis of Lyons (1996), although it runs counter to theresults presented in Dufour and Engle (2000). A further point to note is that, whilethe size of the asymmetric information effect is inversely related to volume, thesignificance of the trade variables is greater in high-volume periods. These resultscontrast with those of Foster and Viswanathan (1993) who demonstrate a positiverelationship between adverse selection costs and trading volume for a sample ofNYSE equities and reject the models of Admati and Pfleiderer (1988) and Fosterand Viswanathan (1990).

Hence, our results suggest that high volume periods are characterised by aconcentration of liquidity trades and increased competition between informedagents, with these effects reducing the price impact of a trade. Fig. 1 graphicallyillustrates the VAR results, plotting the quote impulse response functions (alongwith 95% confidence intervals) for the entire trading day data set plus the 8 to 10GMT and 16 to 18 GMT subsamples in panels (a) through (c) respectively. Thesetwo subsamples were chosen as they are those with the lowest and highest averagespread respectively. The individual panels of Fig. 1 also demonstrate variation inD2000-2 depth across the trading day through variation in theimmediate responseof the midquote to a trade. In panel (c) this is about five times as large as thecorresponding value from panel (b)and thus peak trading periods are characterisedby greater clustering of limit orders around the inside spread.

Based on the VAR results, the return variance decompositions are presented inthe final three columns of Table 2. Across the entire trading day, 41% of thepermanent return variance is attributable to order flow (with a bootstrappedstandard error for this estimate of 0.015). Comparable figures are contained inHasbrouck (1991b) who reports an average value of 33% for a sample of U.S.equities and de Jong et al. (1995) who analyse a sample of French stocks traded onthe Paris Bourse and report an average trade correlated component of 40%. Henceour results imply that the information content of USD/DEM order flow onD2000-2 is of the same magnitude as that on equities markets. Examination of thelast two columns of Table 2 shows that the permanent component accounts foronly one quarter of all return variation such that the information contained in order

10flow contributes one tenth of total return variance. Of course, the distinctionbetween the size of trade correlated component, at 40%, and the 10% number fromthe previous sentence is due to high-frequency, transitory return variation reducingthe latter.

The variance decompositions for the six trading day subsamples are alsopresented. While the VAR results showed that the information content of a singletrade was inversely related to the level of overall traded volume, the variance

10Note that if one performs the permanent / transitory decomposition on data from the 18–6 GMTperiod only, the ratio of permanent to total return variation is less than 0.05. Thus, during this intervalreturns are almost entirely ‘noise’ rather than ‘signal’. It is for this reason that we omit these data fromour econometric analysis.


decompositions show that the information contained in order flow as a whole ispositively related to volume. The ratio of the trade-correlated component to the

2variance of total permanent quote variation (s /s ) is greatest during episodes ofwx v

high traded volume. Furthermore, the proportion of all return variation explainedby order flow information and the size of the permanent component have apositive correlation with volume.

Again, these results can be reconciled with the predictions from theoreticalmodels of intraday variation in transaction activity and price formation. Admatiand Pfleiderer (1988) imply that, in high volume intervals, return variance shouldbe high and prices more informative. In Section 2 we discussed thenegativecorrelation between total traded volume and volatility arguing that the lack oforder book depth in low trading periods was behind this result. However, thevariance decompositions show that both permanent return variation and theinformation entering prices through order flow increase with volume. Aggregatetrading activity hence contributes a greater proportion of all information in peaktrading periods in line with Admati and Pfleiderer (1988).

The prior results provide evidence that the information content of a trade andthat of aggregate order flow are strongly linked to market activity via time of day.Our final set of estimations refine this analysis. The relevant theoretical modelspredict such patterns due to clustering of liquidity trading in equilibrium. Hence,empirical identification of times of concentration in uninformed activity shouldcorroborate the prior results. Given the assumptions on order placement strategymade in Section 3, a concentration of noninformation based trade should beassociated with a high level of limit order placement (the informed preferring totrade by market order due to the execution certainty it yields.)

Data from such episodes should hence display small price impacts from trades.This is examined using the arranged VAR technique discussed in Section 3.3. Boththe first lag of the number and aggregate size of all limit orders outstanding areused as the variable governing the non-linearities in the VAR (z ).t

Fig. 2 presents the moving window OLS estimates of the asymmetric in-Pformation coefficients from the VAR (i.e.o b ) after arranging the data by thei50 i

aggregate limit order size (panel(a)) andnumbers of orders outstanding (panel (b))respectively. These estimates are from an 8 lag VAR using the 6–18 GMTD2000-2 sample. Thex-axis labels in the panels give the sorted observationnumber of the final included data point.

Both panels of Fig. 2 strongly suggest that the asymmetric information problemis lower whenz is higher i.e. in periods of high limit order placement. Conversely,t

when either the number or size of outstanding limits is low, the impact of anunexpected trade on the midquote is large. The relationship is clearly non-linear,however, with the sum of asymmetric information coefficients declining sharplyover the first 15,000 sorted observations and less rapidly thereafter. Nonetheless,the graphs provide evidence consistent with Admati and Pfleiderer (1988) and the‘hot-potato’ hypothesis of Lyons (1996).


Fig. 2. Sum of asymmetric information coefficients from the arranged VAR specifications. Notes: thesePplots show the sum of the asymmetric information coefficients (i.e.o b ) from arranged VARi51 i

estimations, as discussed in Section 3.3. In panel (a) the data are arranged by aggregate order book size(i.e. the sum across limit buy and sell sides of order quantity outstanding in $m) and in panel (b) thedata have been arranged by aggregate number of order outstanding. The values on thex-axis give thenumber of the last observation included in the subsample upon which the VAR was estimated.

We have tested the stability of the VAR coefficients in a number of ways. Forexample, we have estimated extended versions of Eqs. (1) and (2) whereright-hand side variables are interacted with dummies for low, medium and highlevels of liquidity supply. Formal tests for parameter stability confirm the visual


evidence in Fig. 2, yielding strong evidence of structural change in the VARequations related to the level of liquidity supply.

Finally, it is clear that these arranged VAR results tell essentially the same storyas do our results from time-of-day subsamples of the data. When D2000-2 is busy,price impacts of trading tend to be smaller. We have included them here as we feelthat time-of-day effects, in themselves, are uninteresting—the interesting issue iswhat economic phenomenon generates them. We feel that our result relating thesize of information effects to the rate of liquidity supply is a step towardsunderstanding the underlying economics.

5 . Conclusion

This paper analyses trades and quotes from the electronically brokered segmentof the spot USD/DEM market. Specifically, we examine whether this market canbe characterised as subject to information asymmetries. Using the technologyintroduced by Hasbrouck (1991a) and Hasbrouck (1991b) we find that tradesdocarry information. Roughly 60% of the bid-offer spread in our data can be relatedto the asymmetric information problem. Further, around 40% of the variation inthe fundamental price is shown to be order flow driven, a proportion which iscomparable to those found in studies of equity markets.

We uncover clear intra-day variation in the information content of a trade andthe total information content of order flow. In line with the theoretical predictions,in high volume/ liquidity intervals the information content of a single trade is lowwhile the share of information entering the midquote through order flow is high. Amore refined analysis of the relationship between the price impact of trades andmarket activity is also presented. Similar to the time-of-day analysis, results froman arranged VAR estimation indicate that the asymmetric information coefficientsvary systematically with market liquidity.

There are several possible extensions to the current study. First, a more carefulanalysis of the non-linearities in the VAR structure may prove to be fruitful.Parameterising the nonlinearities would likely yield insights into the determinationof price impacts from trade. Another interesting area of research would be toexamine how transaction activity affects subsequent liquidity supply throughanalysis of the excess demand and supply schedules for currency. Furthermore,econometric analysis of individual limit order entry and execution would providestronger evidence on the order placement strategy of dealers and how it reacts totrading activity and quote volatility. Such issues are currently under investigation.Finally, as we have mentioned above, D2000-2 is one of several venues forinter-dealer foreign exchange trade. It would be interesting to evaluate how thesevenues interact (in terms of trading activity and quoting behaviour). This wouldgive a clearer picture of when and where informed trade was trading price.


A cknowledgements

I would like to acknowledge the financial support of the Financial MarketsGroup at LSE and the ESRC (UK). I am grateful to Jon Danielsson, CharlesGoodhart, Roger Huang, Rich Lyons, Carol Osler, Paolo Vitale, seminar particip-ants at LSE and the City University Business School, three anonymous refereesand the Editor, Andrew Rose, for helpful comments. Thanks also to Reuters plc forproviding the data employed in this study. All remaining errors are my own.

R eferences

Admati, A., Pfleiderer, P., 1988. A theory of intra-day patterns: volume and price variability. Review ofFinancial Studies 1, 3–40.

Chakrabarti, R., 2000. Just another day in the inter-bank foreign exchange market. Journal of FinancialEconomics 56, 29–64.

¨de Jong, F., Nijman, T., Roell, A., 1995. A comparison of the cost of trading French shares on the ParisBourse and on SEAQ International. European Economic Review 39, 1277–1301.

Dufour, A., Engle, R., 2000. Time and the price impact of a trade. Journal of Finance 55, 2467–2498.Evans, M., 2001. FX Trading and Exchange Rate Dynamics. Working Paper, Department of

Economics, Georgetown University.Evans, M., Lyons, R., 2001. Order flow and exchange rate dynamics. Journal of Political Economy 110,

170–180.Foster, D., Viswanathan, S., 1990. A theory of the interday variations in volume, variance and trading

costs in securities markets. Review of Financial Studies 3, 593–624.Foster, D., Viswanathan, S., 1993. Variations in trading volume, return volatility and trading costs:

evidence on recent price formation models. Journal of Finance 48 (1), 187–211.Glosten, L., Milgrom, P., 1985. Bid, ask and transaction prices in a specialist market with

heterogeneously informed traders. Journal of Financial Economics 14, 71–100.Hamao, Y., Hasbrouck, J., 1995. Securities trading in the absence of dealers: trades and quotes on the

Tokyo Stock Exchange. Review of Financial Studies 8 (3), 849–878.Harris, L., 1998. Optimal dynamic order submission strategies in some stylized trading problems.

Financial Markets, Institutions and Instruments 7 (2).Hasbrouck, J., 1991a. Measuring the information content of stock trades. Journal of Finance 46 (1),

179–206.Hasbrouck, J., 1991b. The summary informativeness of stock trades: an econometric analysis. Review

of Financial Studies 4 (3), 571–595.Huang, R., Stoll, H., 1997. The components of the bid-ask spread: a general approach. Review of

Financial Studies 10, 995–1034.Lyons, R., 1995. Tests of microstructural hypotheses in the foreign exchange market. Journal of

Financial Economics 39, 321–351.Lyons, R., 1996. Foreign exchange volume: sound and fury signifying nothing? In: Frankel, J., Galli,

G., Giovannini, A. (Eds.), The Microstructure of Foreign Exchange Markets. University of ChicagoPress.

Massa, M., Simonov, A., 2001. Reputation and inter-dealer trading: a microstructure analysis of thetreasury bond market. Working Paper, INSEAD.

Meese, R., Rogoff, K., 1983. Empirical exchange rate models of the seventies: do they fit out ofsample? Journal of International Economics 14, 3–24.


Naranjo, A., Nimalendran, M., 2000. Government intervention and adverse selection costs in foreignexchange markets. Review of Financial Studies 13, 453–477.

Peiers, B., 1997. Informed traders, intervention, and price leadership: a deeper view of themicrostructure of the foreign exchange market. Journal of Finance 52 (4), 1589–1614.

Perraudin, W., Vitale, P., 1996. Interdealer trade and information flows in the foreign exchange market.In: Frankel, J., Galli, G., Giovannini, A. (Eds.), The Microstructure of Foreign Exchange Markets.University of Chicago Press.

Yao, J., 1998. Market making in the interbank foreign exchange market. New York University SalomonCenter Working Paper[S-98-3.

Date post:	01-Nov-2016
Category:	Documents
Upload:	richard-payne
View:	213 times
Download:	0 times

Informed trade in spot foreign exchange markets: an empirical investigation

Documents