1

Village

DefiningWhatTransparencyMeansinthe

WholesaleForeignExchangeMarket

July2017

2

Contents:

AbouttheTransparencyTaskForce 3

ExecutiveSummary 4

Rationale 6

MarketStructureRecap 7

HiddenFXCosts 10

ThecaseforPortfolioCompression 17

ElectronicTrading:Anewtimescale18

HowMuchDoFXTransactionsCost?25

RetailFX;thesignificanceforpensionsavers28

Recommendations30

3

AbouttheTransparencyTaskForce

The Transparency Task Force (TTF) is a campaigning community, dedicated to driving up

levels of transparency in financial services, right around theworld. It believes that higher

levels of transparency are apre-requisite for fairer, safer andmoreefficientmarkets that

willdeliverbettervalueformoneyandbetteroutcomestotheconsumer.

Furthermore, because of the correlation between transparency, truthfulness and

trustworthiness,theTTFexpectsitsworktoimprovethereputationofthefinancialservices

sector.

TheTTFseekstooperateinacollaborative,collegiateandconsensus-buildingway;focusing

on solutions,notblame. Ithas over160volunteersorganised into9 teams.Each team is

workingonseparatecampaigninitiatives.TheTTF’sForeignExchangeTeamisledbyAndrew

Woolmer,CoFounderandCEOofNewChangeFX.

http://www.transparencytaskforce.org/

LeadAuthor:XavierPorterfieldCFA,HeadofResearchatNewchangeFX

SupportingAuthor:JessicaBilcock,GovernmentRelations,TransferWise

4

ExecutiveSummary

Thesecondandfinal instalmentofanewGlobalFXCode,designed“topromotearobust,

fair, liquid,open, andappropriately transparentmarket”hasnowbeenpublished.MiFID2

and PRIIPs come into force in January 2018, requiring fiduciary managers and service

providerstoaddresstheirFXprocesses.

TheGlobalCodeisavoluntary,principlesbasedsetofstandards.Thepurposeofthecodeis

toprovideaharmonised,globalstandardofbestpractice intheglobalFXmarket. Indeed,

thefirstpartofthecode,releasedinMay2016offeredusefulexamplesandcasestudies.

The Global Code came into being as a response to political pressure following various

reviewsandenquiriesintoForeignExchangepractisesinnumerousjurisdictionsaroundthe

world. Due to its size, (over one third of theworld’s global FX volumes are exchanged in

London)theUKgovernment’sFairandEffectiveMarketsReviewplayedakeyroleinframing

thediscussionoftheissuessurroundingthewholesaleFXmarket.

Achievingagreementonaglobalsetofstandardsisalaudableachievement.However,the

releaseoftheglobalcodedoesnotaltertherealitythatOTCmarketsremainfundamentally

opaque. Asetof standardscannotchange that. Theareasof thecodewhichcaused the

most controversy were related to issues related to trading ahead of clients (to source

5

liquidityforclients)andthepractiseof“lastlook”.Intheend,nofinalsetofprinciplesthat

governthesepractiseswasagreed.

ToachievetransparencyinForeignExchange,clientsmustbeabletobeabletodefinewho

theyaretradingwith,preciselywhentheyareexposingatradetothemarket,andatwhat

absolutecostthetradehasbeendone.

This paper seeks to explorewhere the FXmarket remainsopaque, andhow transparency

canbeachievedinthesethreecriticalareas.

TheTTFrecommends:

- CustodiansnottransactFXonaprincipalbasis.

- AllFXtransactionsshouldcomewithreliabletimestamps–thisistimeimmediately

priortotheorderarrivingwiththetradingdeskthatexecutestheorder.

- Thearrivalprice formeasuring themarket impactofpoint in timeFIX transactions

should take the market mid-rate available at the moment the fixing order was

transmittedtoanintermediary.

- Transaction costs should be measured against objective, independent data that

reflectsthebestmarketpriceavailableinthemarketwhentheorderwassubmitted

totheirintermediary

6

- For retail transactions, these transaction costs should be disclosed in full to the

consumer, including the revenue percentage included in the provider’s reference

rateontheday.

Rationale

The foreign exchange market is a delocalised, global market where foreign exchange

transactionscanoccuraroundtheclock.AclientsittinginCaliforniacantradewithamarket

maker in London for a fundwhich is registered inHongKong. Theperception that theFX

marketisriggedpromptedaco-ordinatedresponsefromregulatorsandmarketparticipants,

cognisantthatunlessactionwastakeninconcert,poorbehaviourwouldsimplymigrateto

jurisdictionsofferinglowerregulatoryconstraints.

Lack of transparency in OTC markets can have a number of undesirable effects. These

includeamisallocationofcapital. IfthecostsofparticipatingintheFXmarketareopaque,

orifthemarketisperceivedasbeingrigged,participationwillbediscouraged.Thiscanlead

to a casino image of the FX market where participation is seen as a form of gambling ,

leavingcurrencyexposuresunderhedged.Thisunderhedgingmeansthatendinvestorsare

exposedtounnecessaryorsub-optimalrisks,allofwhichincreasethecostburdenonsociety

asawhole.

Removingopacitycouldencouragegreatermarketparticipationandincreasetheefficiency

of FX market participation. The global FX code represents a major achievement in

7

international co-operation, but unfortunately, a number of the features of the FXmarket

whichcreateopacityhavenotbeendealtwith.Thispaperseekstoremindregulatorsand

stakeholdersthattheworktowardsasuitableleveloftransparencyisnotcomplete.

MarketStructureRecap

TheFXmarket isdecentralised. Liquidity isdisaggregatedovermanydifferent venuesand

counterparties.Moreover, FX is tradedonabilateral rather thananexchangebasis. This

means that every FX price is customised for each client. The ‘going rate’ in the Foreign

Exchangemarketdependsentirelyonwhoisaskingthepriceofwhom.

Historically, the wholesale FX market was two-tiered. An exclusive ‘interdealer’ market

segmentallowedbankstosourceliquidityfromeachotherandadealer-customersegment

allowedcustomerstosourceliquidityfromthemarket-makingbanksinturn.

The interdealer segment has now lost its exclusivity. In 2005 interdealer prices and data

finallybecameavailabletothecustomersegmentforthefirsttime.Coupledtochangesin

thecredit structure through theevolutionofPrimeBrokerage, thisgavenon-bankmarket

participantstheabilitytotradeonthesamepricesasbanksand,moresignificantly,tomake

pricestoo.

8

ThemajorityofFXvolumeisnowconductedonmarketvenueswhereliquidityflowsinone

direction only: frommarketmaker to customer. This reflects the legacy of a two-tiered

systeminwhichcustomerscontinueto‘take’pricesandpaymarketmakers,despiteowning

liquidity.

With liquidity disaggregated into numerous pockets or pools,marketmakers provide the

necessaryintermediationtosourceliquidityforclients.Marketmakersdonottypicallyearn

explicit fees, instead, they seek tomake profits by earning a bid-ask spread. This spread

represents a risk transfer fee. The client closes their risk by transferring it to themarket

maker–andpaysafee.

Whenmarketparticipantsfaceeachotherasprincipaltheycompeteoverthetermsofthe

deal. The interests of the principals are diametrically opposed. Market makers seek to

maximisethespreadtheycanearn,whilecustomerstrytominimizethespread.

Inadditiontospread,participantscompeteovertheinformationcontentofthetrade,which

can be much more valuable. FX prices are formed before a transaction occurs, via

expressionsofinterest.Thepriceshownisjustanindication,andthepricecanbewithdrawn

if it becomes unfavourable to themarketmaker showing the price. But in the process of

showingapricetoapotentialbuyer, themarketmakerhasgleanedvaluable information

aboutmarketinterestatthatprice.Marketmakersseektoavoidfallingintothetrapofthe

buyer’scurse,otherwiseknownasadverseselection-owningacurrencythatisworthless

thantheypricetheyjustpaidforit.

9

Changes to the regulatory environment, particularly with respect to the capital charge

appliedtoriskweightedassets,havediminishedtheappetiteofmarketmakerstoallocate

capital tomarketmaking.Thishasmeant that traditionalmarketmakers (thebanks)have

been less willing or less able (or both) to warehouse risk. This has led to much shorter

inventoryholdingperiodsandasignificantincreasein‘matchedprinciple’trading.

Matched principle trading entails a market maker effectively white labelling the liquidity

solution of another provider. Typically they will add a mark-up to someone else’s rates,

ratherthanactuallyearningthebidaskspreadbytakingprincipalrisk.

Much ofwhatwe callmarketmaking actuallymore closely resembles exploitation of risk

free arbitrage. Market making institutions seek to preserve their informational edge,

because the source of their profits is not from taking principal risk. Instead, profits come

frombeingbetter informed than their clients– and charging risk transfer fees for riskless

arbitrage.

Thenewrulesonriskweightedassetshaveencouragedthosewantingtotakemarketriskto

investinnon-bankmarketmakinghedgefundsratherthanallocateriskdirectlytoin-house

market making. Unsurprisingly, the new entrants into the market making space are less

constrainedinhowtheyusetheirowncapitalthanthebanks.Thesefirmsarenowstepping

intorolesthatonceweretheuniquepreserveoftheinterdealerbanks.

10

Aswediscussedabove,inprincipaltrading,marketparticipantstradeontheirownaccount,

puttingtheirowncapitalatrisk.Principalactorsseektobuyandsellontermsthataremost

favourabletothemselves.Inamatchedprincipletrade,afacilitatorbringstogetherabuyer

andaselleronanagencybasis,withouttakingtheothersideofeitherparty’s trade. The

transaction is completed simultaneously with the facilitator’s remuneration remaining

independentoftherateatwhichthetransactionoccurred.

However, as the recent Fair and EffectiveMarkets review concluded, there are instances

whenthedistinctionbetweenagencyandprincipalareblurred.

HidingFXCosts:

1. AgentstradingasPrincipals

Despite a spate of lawsuits highlighting the problem, a significant number of investment

firmshavenegotiatedarrangementswithcustodians,grantingthecustodianamonopolyon

theFXtransactionsthatoriginatefromthefund.Thisobligestheunderlyingfundinvestors

toacceptthecustodian’sFXratesevenwhentheyarenotcompetitive.

A Russell Research paper “It’s time for more choice in FX” published in 2004 raised

awarenessof theextent towhich FX transactions couldbe costing investors. Theirwork,

basedonthousandsoftrades,showedthatthemajorityofcustodialFXtradeswereheavily

skewedtotheworstratesoftheday.

11

The issue came to a head in 2009 when the largest pension fund in the United States,

Calpers, sued their custodian. Theheart of theproblem is that custodiansmay choose to

executeFXtradesonaprincipalbasis,puttingtheirowninterestsindirectcompetitionwith

theinterestsoftheunderlyinginvestorswhoseassetstheyhold.

Investors caneasily be confusedas towho they aredealingwith, andonwhatbasis.Are

their funds traded with someone looking after their interests, under the protection that

agencylawsprovide,oraretheytradingwithsomeoneonacompetitivebasis,asaprincipal,

withouttheprotectionofagency?

The evidence from years of Transaction Cost Analysis suggests fund investors are much

betterservedwhenfiduciariesarenotgiventheopportunitytotradeagainsttheinterestof

thecustomerstheyserve.

TheFairandEffectiveMarketsReview(2015)callsfortheGlobalFXCodeto“setstandards

for the treatment of clients and counterparties. This section of the code should address

issuessuchasthepreventionandmanagementofconflictsofinterest,especiallyconcerning

mixedprincipalandagentroles”

Andyet,custodialFXcontinuestobeofferedto investmentfundclientsonaprincipalrisk

basis,andtheGlobalFXCodedidnotaddressthisissue.

12

2.TimeStamps

A study conducted by researchers at Brandeis International Business School andWilliams

College, (Olser, Savaser, Nguyen 2012) found that asset managers may be choosing to

outsourcetheirFXtradinginorderto“shroud”costinformationfromunderlyinginvestors.

Therationaleforthisstrategyisthatunderlyingfundinvestorsmaynotnoticetheefficiency

gains from negotiating currency deals individually. Improved performance might be

accredited to other factors. On the other hand, negotiating deals individually requires a

trader, technology and trade processing staff, all of which are likely to result in higher

running costs that are much easier to identify. Transaction efficiency is hard to isolate.

Costsaremuchmorevisible.

Olser,SavaserandNguyenstudiedthecompletetradingrecordofamid-sizeglobalcustody

bankwhichincluded70,000transactionsin25currencies.Theyfoundthattheaveragecost

of non-negotiated FX trades was 19 basis points, well above the 2-3 basis points clients

mightexpecthadtheyexecutedonanegotiatedbasis.

When FX orders are “shrouded”, prices are set relative to the high and low for the day.

Becauseacustomer’slossisaprincipalcounterparty’sgain,thedistributionofcustomerfills

areheavily skewedto theworstpricesof theday.Asevidencedby research fromRussell,

(2004opcit,revisitedin2010)andRecordCurrencyManagement(2011).

13

ThemovementofassetflowsthatnecessitateFXdealsareknownlongbeforetheFXtradeis

done. CustodialFXdealersareabletoacquiretheinventorytheyrequirediscretely,filling

clientordersbeforetheymaketheirclientaprice.Theyarethenwellplacedtoaddamark

uptotheirownacquisitioncosttomakesuretheyalwaysmakemoneyonthetrade.This

effectivelyturnstheFXtradeintoarisklessarbitragewhichdoesnotjustifytheearningofa

risktransferfee.

Theseeffectsareparticularlyamplifiedinrestrictedcurrencies,wherethecustodiancannot

by lawdoanFXdealwithoutanunderlyingassettransaction. Thecustodianwill insiston

theassetdealsettlingbeforetheFXhedgeisdone,whichmeansthatinvestorsoftenwait4

days before their FX is hedged. During this time the custodian can opportunistically pre-

hedgethedeal–andguaranteethemselvesaprofit.

Abuse isdifficult tospot.Custodial tradereportscontainnoreferencetothetimenorthe

relevantmid-market ratewhen the trade took place. Thismakes it extremely difficult for

assetmanagerstoidentifytheirrealisedFXexecutioncosts.

Toencouragetransparency,theFairandEffectiveMarketsreviewcalledformandatorytime

stampingonallFXtrades. Thissimplestephasbeen ignored intheGlobalCode,allowing

custodianstocontinuetheirpractiseofhidingtimestamps.Obligingcustodianstorelease

timestampsofnon-basecurrencyequityandbonddealswouldofcoursebesimpleenough

–andenableFXcostmeasurement.

14

EvenwitheffectiveTime-Stamping,automated,non-negotiatedFXdealsremainvulnerable

tootherformsofcostshrouding.

3.The4pmFix

TheWMR Fix is a valuation tool used to value trillions of dollars in common investment

benchmarks.Because theunderlyingequity andbondbenchmarkprovidersuse theWMR

Fixtovaluethebenchmarkportfolio,passivefundmanagershaveanincentivetotrytopeg

their FX transactions to the benchmark rate. This serves to minimize the tracking error

betweenthebenchmarkvaluationrate,andtherealizedexecutionratesoftheirportfolios.

RespondingtoconcernsabouttheintegrityofFXBenchmarks,theFinancialStabilityBoard

taskedaspecialworkinggrouptoreviewFXBenchmarks.TheirworkfocusedontheWMR

4pmfix,andtheECBfixing,thetwomostwidelyusedbenchmarks.

The working group final report (September 2014) concluded, “it is the incentive and

opportunityforimpropertradingbehaviourofmarketparticipantsaroundthefix,morethan

the methodology for computing the fix (although the two interact), which could lead to

potentialadverseoutcomesforclients.”

TheEuropeanCentralBanknowactivelydiscouragestheuseofitsownECBreferencerate

fortransactionpurposes.

15

However, the FSB working group did not review the cost to investors of using these

benchmarks.

Independent research conducted by numerous firms and institutions (including but not

limitedtoNewChangeCurrencyConsultants,PragmaSecurities,NewCityInitiative)reveals

thatusingtheWMRFixishighlyinefficient.

As fixing orders must be submitted at least 30 minutes before the fix, market making

institutionscan(andinmanycasesmust)beginpre-hedginglargeorderswellinadvanceof

theFix itself. Speculators canof course see thisactivity,andareable to jumpaheadand

frontrunit,whichcreatesskewinpricing,pushingthepriceagainsttheusersoftheFix.This

meansthatthecosttocustomersisoftenmanytimeshigherthantheFixitselfshows–but

they cannot see this cost because they have achieved the fixing price, and are only ever

shownthecostofexecutionwithin the fixingwindow. The fixingwindowopensafter the

skewhasbeenachieved.

Usersof the4pm fixareparticularly vulnerable to these skewcostsatperiodends,when

largeportfoliorebalancesthatmatchpopularinvestmentbenchmarksoccur.Theamountof

moneytrackingparticularbenchmarksensuresthatthedirectionofthemarketforthefixis

predictable. Thiserror is thencombinedwith thepopularmisconception thatFXmarkets

arehighlyliquid.InfactFXmarketsarehighlyilliquid,withamaximumofUSD14.4million

dollarsasecondbeingtradedinEURUSDgloballyatthebusiesttimeofday.

16

In lieu of time stamping trades, custodians and managers often pretend to offer

transparencyfortheirtransactionsbydeclaringthattheywillconvertcurrencyattheWMR

fixprice.

Becausemarket impact canonly be seen in the context of the trading activity immediate

beforeandafterthetrade,marketimpactistoallintentsandpurposescompletelyinvisible

totheunderlyinginvestorsinafund.

The choice to use the 4 PM fix is simply another shrouding method, chosen to hide

transactioncosts.Whatismore,thisparticularshroudingmethodismassivelyexpensive.

NewChangeFXconductedastudycomparingexecutionratesusingatimeweightedaverage

price and the 4 pm fix. Assuming passive investors will normally be positioned in the

directionofthefix,thestudyfoundthatusingtheFixforEURUSDtradesresulted inanet

costtoNAVof3.6%overayearcomparedtotransactingatatimeweightedaverageprice.

Comparing this result to realised trades in EURUSD of UK and US based assetmanagers,

NewchangeFXfoundthat investorswere incurringbetween60%and70%of thesecosts -

that is about $23,000 permillion. These are costs that are unreported, as the timestamp

usedistheopenofthefixingwindow-andwillcontinuetobeunreportedunderMiFID2.

17

When a fund had no contractual obligation to use the 4pm Fix for some index tracking

purpose,costshroudingmaybeafactorinthechoicetousetheFix.

TheCaseforPortfolioCompression

Costshroudingisaspecialcaseofopacity,butuseoftheFIXalsopresentsanotherproblem.

Itisnotefficient.ThespecialworkinggroupmandatedbytheFSBtoinvestigatetheuseof

FXbenchmarks“supportsthedevelopmentofindustry-ledinitiativestocreateindependent

nettingandexecutionfacilitiesfortransactingfixorders”(Section7,Point6oftheForeign

ExchangeBenchmarksFinalReport)

Intheirreview,theFSBworkinggroupfoundthatclientswereusingtheFIXtotrade,when

nettingmight be amore cost effective solution. The problem is that appropriate netting

facilitiesdonotexist,orarenotwidelysupported.

Marketmakersliveonflow.Nettingoftradesreducesthevolumeofflowthatisavailableto

marketmakers.A recent studybyNewchangeFXbasedon theFXexposuresof just4UK

pensionfundsfoundthatnettingcouldreducetheircombinedFXcostsby60%.

Byexecutingeach transaction individually, rather thannettingdown toproducea smaller

market exposure, FX volumes are larger than they need to be. Portfolio compression can

offer substantial cost savings. The problem is that market makers have no incentive to

support or create external netting facilities. This would seem to offer an opportunity for

18

disruptivenewentrants.ButheretheFXmarketpresentsastructuralobstacle.Becausethe

dominantFXmarketmakersarealsodominant inclearingandsettlement it isdifficult for

moreefficientmethodsofexecutionsuchasexternalnettingfacilitiestoemerge.

Despite the vested interest of incumbents, a number of portfolio compression initiatives

have been launched. For the time being these services are only available to banks, but

LMRKTSaportfoliocompressionserviceprovidefornonCLScurrenciesdoesplantoopenup

to some buyside firms. OTCmarkets remain dependent on credit, which creates a tiered

hierarchyofwhohasaccesstotheseservices.Ineffect,unlessauniversalaccessapproach

can be found, portfolio compression can become another differentiator, re-introducing a

softerformoftwotieredmarketwherethebestnamescanbenefitfromcompression,while

othersmaynot.

ElectronicTrading-Anewtimescale

Electronic trading has become the dominant medium for conducting Foreign Exchange,

accountingforupto65%ofaveragedailyvolumes.Thedevelopmentofelectronictrading

hasintroducedanewdimensiontohowmarketmakerscangainanadvantagesoverprice

takers.Timeismoney!

19

Electronictradingandvoicetradingdonotoccurinthesametimescale.Reportedresponse

timesfromanumberofECNplatformsareintheorderof300microseconds.

However,itisextremelyunlikelyacustomerwouldbeabletocompleteatransactionwitha

marketmakerwithin this time frame. The response time from amarketmakermight be

anywherebetween10and1000timesslower.

Price slippage occurs when the actual transaction price differs from the price when the

decision to tradewasmade. Firmpricing inForeignExchange is still rare.Themajorityof

prices are displayed as an indication of interest, which can bewithdrawn at amoment’s

notice. If thepricecanbewithdrawn,customerscanwonderwhetherthepricewasreally

thereatall.Thisgivesrisetoaphenomenonknownasphantomliquidity.

One of the areas that has caused controversy in the FXmarket is how banks have been

applying last look. Last lookensures the client is alwayswrongand that theprice-maker

securesarisk-freeprofit.Itisanembeddedcomponentoftrading,andthankstoveryweak

handling of the issue by the sponsors of the Global Code, it remains so. The issues are

complicated and the Global Code simply ensures that last look can be given more

respectableattirethanusedtobethecase.

Theissuesanddefinitionsareasfollows:

20

Latencyslippageiscausedbythetimeittakesforasignaltotraveloveranetwork.

Latencyslippageshouldbesymmetrical.Afterallitissimplytheresultofasignalbeingout

ofdatebythetimetherecipientof thesignal receivesthemessage. Amarketmakerhas

updatedaprice,buttheupdatedpriceisnotvisibletothecustomeryet,whoisattempting

totradeonapricethatisstale.

LastLookslippageoccurswhenamarketmakerhasleftpricesinthemarketthathavebeen

supersededby events. This sounds like latency, but it is not about the time it takes for a

signal topropagate througha system. It is aboutwhether themarkethasmovedand the

marketmakerhasbeenslowtoadjusttheirprices.

Last look addresses a problem where the market maker has been slow to respond to a

changeinmarketconditions,asopposedtoalatencyissuewhichdescribesthetimeittakes

signalstotraveloveranetwork.

The problem arises, and the opportunity for abuse occurs because latency is reported in

differentways.Itmaybeusedtorefertothetimeittakesavenuetoshowapriceupdate

fromliquidityproviders,oritmayrefertothetimeittakesforaclienttotradeonpricethat

hasbeendisplayed.

21

LastLook–AdverseSelection

Lastlookwasintroducedbymarketmakersasameanstoprotectthemselvesfromlatency

arbitrage. Prices change in response to new information.Marketmakers will lose if they

make prices to a better informed trader ( a traderwho can respond to new information

quickerthanthemarketmaker),andwinwhentheytradewithcustomerswhoarelesswell

informed.

A price that updates too slowly is vulnerable to being picked off by a trading algorithm

systemthatisabletoreactfaster.Thiscreatesadverseselectionriskforthemarketmaker-

wherethemarketmakermightenduplong(owning)acurrencythathasfalleninvalue.

Last look has created controversy because some market makers have applied last look

asymmetrically,whichistosayunfairly.

Considerthefollowingprocessofhowanorderistransactedonapricestream:

- Clientopensalivestream.ThelivestreamisaseriesofFiXQuoteMessageswhich

arecontinuallycancelledandupdated.

- Clienthits aprice and indoing so instantly sendsanordermessage to themarket

maker.

- Themarketmakeropensaholdwindowfortheclient.

- Attheendoftheholdwindow,themarketmakerrunsatolerancecheck.

22

- Ifthedeal iswithinthetolerancecheck,aconfirmation(AC)messageissent, ifnot

thenarejectionmessageissent(NAC).

Thisprocessapplies toall clients. The tolerancecheck iseitherdefined inbasispointsor

USD terms. ByusingUSD terms, the clientwill notbe rejectedon small deals,which can

happen if basis point variances are used. The holdwindow is continually updated as the

machine learns how a client behaves. Both variables are set on a bespoke basis per

client.Thetolerancelevelcanbe'reversed'wherebythebankactuallyacceptsacertainlevel

oflosspertransactionbeforeitisrejected.

Aswenotedearlier,responsetimesarebasedonthetimeittakesasignaltotravelacrossa

system.Thissignalitselfhasvariability-itfollowsasinewave.Itisverydifficulttoachieve

full transparencyonsignal lengthsbecauseof thisvariability,butclearly,movingresponse

timearbitrarily,suchasexplainingtoaclientthereasonforarejectwasbecausethesystem

wasrunningveryslowlytodayareclearlyareasthatarevulnerabletoabuse.

The Request for Quote (RFQ) protocol is typically handled much the same way, but the

openingprocedurestartswiththeclientpullinginaprice,ratherthanreceivingpricesthat

are “pushed” via a stream.Last look canbeemployedonboth streaming and request for

quoteprices.

Whereasapplying last lookonstreamingpricesmightbejustifiedtohandle latency issues,

applyinglastlookonRFQissimplybanksprotectingthemselvesfromthepossibilityofbeing

23

pickedoffina‘drivebyshooting’asliquiditygetsconsumedinamarketsweep.Thelastto

price in the sweep is likely to be disadvantaged in much the same way that a game of

musicalchairsalwayshasonechairlessthanthenumberofplayers.

Thenewglobalcodereaffirmsmarketmakers retainsolediscretiononwhethera trade is

accepted or not. This means that the Global Code has retreated from encouraging

symmetriclastlook(rejectingbothfavourableandunfavourabletradesequally).Ifthegoal

oftheGlobalCodeistopromoteappropriatetransparency,failingtoinsistonsymmetriclast

lookcanonlybeafailure.

Symmetric last lookallowsthebanktheopportunitytore-pricetothecustomer,basedon

the information the clienthas transferred to thebank. Itwill always create slippage,but

whereas asymmetric Last Look invariably creates negative slippage, applying last look

symmetrically theoretically creates a situationwhere the clientmight experience positive

slippage.Givenascenariowherethecustomer’sintenthasalreadybeentransferredtothe

marketmaker,inpracticetheopportunityforpositivepriceslippageisslim.

Ontheotherhand,asymmetric last looksimplyprovidesamarketmakerwith theoption,

butnottheobligationtotradewithacustomer,afterthecustomerhasfullydisclosedtheir

tradingintentions.Howisthisfair?

TheGlobalCode’sapproachlegitimisesLastLook,whenamoretransparentsolutionwould

befairer.Justasbilateralmarketsarebasedonmarketmakersbeingabletoquotedifferent

24

prices to different people, Last Look is designed so that response times are tailored

individually.Whenthingsaredifferent,itisdifficulttocomparethem.

FirstLook

Electronictradinghasopenedupanewavenueformarketintermediariestogainanunfair

advantage over their clients. By definition, when custodians have a monopoly on the FX

transactionsofafundtheybenefitfrominsideknowledge.

In the last few years there have been at least two instances of financial intermediaries

offeringagency tradingordirectmarketaccess (DMA),whilst simultaneouslyoperatingan

undisclosedprincipalmarketmakingdesk.

Spottingthistypeofabusecanbeverychallenging.InarecentcasetheCFTCwithdrewthe

retail fxbroking licence froma leadingFXplatform, FXCM.The firmdidnotdisclose their

interest in a market making firm that clients were trading with via their platform,

misrepresenting to clients that it’s “no dealing desk” platform did not have a conflict of

interestwith itsclients.Customerscontinuedtobenefit fromverytightspreads,sothat it

mightappearthatclientswerenotdisadvantagedbyFXCM’sbehaviour.

Earningthespreadwasnotthegoalofthehustle. Thevalueoffirst look isthebenefitof

beingabletoselectivelychoosewhichtradestoprice,andwhichto let flowontooutside

25

marketmakers.Themarketmakingdesk,byhavingfirstlook,wasabletochoosetradesthat

werefavourabletotheir tradingbook.LastLookcanbeappliedretrospectively, increasing

adverseselection.Firstlookforcesclientsintoadverseselectionprospectively.

Untilseniormanagementspottedthebehaviour,employeesofITG,thebrokerdealer,rana

similarfirstlookoperationintheirUSequitiesbusiness.(SECpressrelease12thAugust2015)

ITGand itsaffiliateAlternetSecuritiesagreed topaya$20.4million fine to settlecharges

thattheyoperatedasecrettradingdeskandmisusedconfidentialtradeinformationofusers

oftheirdarkpool.

Inordertospotfirstlookorlastlookabuses,forensictransactioncostanalysisisrequired,to

quantifyfillratiosandposttradedecaywithindividualcounterparties.

HowmuchdoFXtransactionscost?

The early pioneers of FX Transaction Cost Analysis, firms such as Record Currency

Management or Russell Investments faced particular challenges when they attempted to

uncoverthehiddencoststhatpensionfundswerepayingfortheirFXtransactions.

As noted above, custodians do not typically provide time stamps, nor do they provide a

referencetotheactualmarketmid-rateprevailingatthetimethetransactiontookplace.

26

Time stamping is becoming more prevalent and custodians have made improvements to

theiroperations.However, thequestion remains, if everybody receivesa slightlydifferent

price, what should we measure against, to enable comparability across fund managers,

marketmakersandvenues?

Customershavetypicallyaddressedthequestionofslippagebycomparingtheirexecutionto

thepricesthatweremadeavailabletothem.ThismethodseemedtobesupportedbyMiFID

Iwhichproducedarequirementtoputliquidityprovidersintocompetition.

The troublewith this approach is that determining execution costs in thisway is entirely

circular.

For instance, a largeUKassetmanager createdamid-ratedatabase fromBids andOffers

collected from 10 liquidity providers. This ignoredwhether therewas not a better quote

elsewhere,offeredbyacounterpartythattheydidnotdealwith.Choosingtomeasuretheir

costs in this way caused the asset manager to understate their FX transaction costs by

approximately£10millionayear.

Inmethod,thisisonlyalittlebetterthancomparingexecutiontothepricesthatweremade

availabletothefundbythecustodian’sFXdesk,dealingwithamonopolyonthefund’sFX

deals.

27

Fromaninvestor’sperspective,weneedtoknowhowatransactionoccurredrelativetothe

bestavailableprice in themarketasawhole.Thiscanonlybedeterminedbyaggregating

bids and offers across a large cross section of venues that is representative of thewider

market.

Because theFXmarket isdisaggregated, theprices fromsingleaggregatedvenuessuchas

EBS or Reuters do not necessarily indicate the best price available in themarket. Using

singlevenuedataintroducessamplingerror.

Cognizant of the increasingly fragmented nature of the market, European regulators

proposed a transparent framework for reporting transaction costs. The Packaged Retail

InvestmentandInsurancebasedProducts(PRIIPS)requiresfirmstocapturethemarketmid-

rateprevailingatthemomenttheorderwascommunicatedtoathirdpartyforexecution.

Moreover, themarketmid-rate(calledthearrivalprice)mustbeaconsolidatedprice,and

notapricefromasinglecounterpartyorplatform.

Applying and recording a consolidatedmid-price (arrival price) is amajor breakthrough in

costtransparency.

28

ConflictedData

Asinvestorslearnttotheircostin2013,whenliquidityproviderscandirectlyinfluencethe

price that they are measured against, they will have both opportunity and incentive to

manipulatetherate.

Singlesourcedata thatan investorhas tradedonprovidesacircularmeasurement.When

liquidityisfragmented,measuringexecutionfromasinglesourcemeansmeasuringagainsta

much smaller liquidity pool. For instance, if the average volumeof a trading venue is 10

billionadayacrossall currencies,normalisingvolumes for the tradingpairwill reducethe

volumetoabout2.5billioninEURUSD,which,spreadoutoverthetradingday,mightmean

thattheactualliquidityavailableinallcurrenciesduringasingleminuteperiodmightbeas

littleasUSD4million,orevenless.

Measuringexecutionagainstaratefromthesameplatformmeansthatit isverylikelythe

investor ismeasuring the quality of their trade from their own trade, drinking their own

bathwatersotospeak.

RetailFX;thesignificanceforpensionsavers

Forconsumers,it’sevenhardertounderstandhowmuchtheyarebeingcharged.Asnoted

above,custodiansdonottypicallygiveatimestamp,orreferencetoanactualmidmarket

rate, when conducting transactions on behalf of pension funds - making it difficult to

ascertaintheexactamountpensionfundsarepayingfortheirtransactions.

29

Onasmallerscale,thereisasimilarissueataretaillevel.Brokersarenotobligedtoprovide

an independent, non tradeable rate on an ex-ante basis as a reference to the customer

attemptingtotransactinadifferentcurrency.

Right now, banks and bureaux de change are only obliged to provide a ‘’reference rate”

which, under thedefinitionprovided in thePayment ServicesDirective and in the second

Directive, can either be an independent publicly availablemid rate or the provider's rate

offered on the day - which includes their revenue percentage and as such constitutes a

chargetotheconsumer.

Theprovider’srateonthedayisanarbitraryconstruct,setentirelyatthediscretionofthe

PSP, with no regulatory oversight. It is far from an independent yardstick by which a

consumercanunderstandthetotalamountchargedforatransaction- includingtheprofit

percentageimbeddedintherateofferedbytheproviderontheday.

This setup requires an unreasonably high level of financial literacy on the part of the

consumer. Research conducted by YouGov shows that only 10% can understand how to

calculate the charge when presented with a typical banking structure1. On average, high

streetprovidersadd£29.54onatypical£1,000GBP>EURtransaction2inanexchangerate

mark-up inadditiontotransactionfees.Overthecourseof2015,UKconsumersandsmall

1 ResearchconductedbyYouGovsurveyed19,277Europeanadultsbetweenthe8th-22ndFebruary2 Averagerepresentativeofatypical£1,000GBP>EURtransfer,usingtraditionalhighstreetproviders.ResearchconductedbyConsumerIntelligenceinFebruary2017

30

businessespaid£5.6bn in charges imbedded in theexchange rate,despite just20%ofUK

consumersunderstandthattheirproviderimbedsachargeintheexchangerate3.

Recommendations

Thegoaloftransparencyisnottomaketheprovisionofliquidityunprofitable.Rather,itisto

removetheopacitysurrounding liquidityprovisionsothatcustomerscanbesuretheyare

beingtreatedfairly.

As we have seen, the rise of electronic trading provides new ways that information

asymmetries andplainold fashioned cheating canoccur. Thepushof regulation andnew

rulesonthecapitalchargetosupportmarketmakinghasmadeliquidityprovisionlessabout

marketmakers takingprincipal riskandmoreaboutdealersearninga risklessspread. Ina

riskless trade, the economic interest of principles go head to head over information. The

spreadisjustthecherrythatsitsontop.

Identifying and managing conflicts of interest is vital. However, because cost shrouding

strategies are so difficult to spot, and potentially so costly to investors, certain practices

needtobechallenged.

- CustodiansshouldnottransactFXonaprincipalbasis.

3ResearchconductedbyCapitalEconomics,August2016

31

- AllFXtransactionsshouldcomewithreliabletimestamps–orthetimestampshould

betakenfromtheassetsaleorpurchaseandnottheFXdeal.

- Thearrivalprice formeasuring themarket impactofpoint in timeFIX transactions

should take the market mid-rate available at the moment the fixing order was

transmittedtoanintermediary.

- Transaction costs should be measured against objective, independent data that

reflectsthebestmarketpriceavailableinthemarketwhentheorderwassubmitted

totheirintermediary.

- For retail transactions, these transaction costs should be disclosed in full to the

consumer, including the revenue percentage included in the provider’s reference

rateontheday.

Theseproposalsmightbeconsideredsomethingofabigbang,butgreatertransparencywill

result in better risk management, encouraging market participants to trade in the best

interestsoftheircustomers.