The Marketmaker in the Age of the ECN

8/14/2019 The Marketmaker in the Age of the ECN

1/13

JounNer- Or INVEsTMENT MeNeceurvr, Vol. 2, No. 1, (2004), pp. 4-15@ JOIM 2004

www.lorm.com

INSIGHTSTHE MARKET MAKER IN THE AGE OF THE ECN

Valtne H. Wagner*Electronic nading uenues demonstrate an impressiue ability to successfully match trad.eswith high accuracy, low cost, and attrading process, or will there always

Market makers find themselves under attack fromall sides. Bid-offer spreads, the bread and butterof market making, have been reduced by com-petitive forces and relentless pressure from theSecurities and Exchange Commission. Marketsonce serviced exclusively by market makers, are nowchallenged by alternate brokerage arrangements andElectronic Communications Networks (ECNs).Investors voice the widespread sentiment that theywould prefer to live without market makers.Not so fast! It is time for investors to step back andconsider whether market maker services are worthyof such apparent disdain. Aspects of market makingare essential to the smooth functioning of a market,and should be encouraged rather than disparaged.I Tiading forumsEconomists theorize a classical market in terms of alarge number ofsmall participants, in rough balance

*C/O Plexus Group, Inc., 11150 Olympic Blvd., Suite 900,Los Arrgeles, CA90064.

"Insights" features the thoughts and views of the top authoritiesThis section offers unique perspectives from the leading minds in

remarkablebe actiuities

@

from academia and the profession.investment management.

speeds. Are they on track to take ouer tltebetter performed by human beings?

of buyers and sellers, interacting to find the pricesthat create the largest possible volume of buyersand sellers. Never does one buyer or seller havedominant infuence over the market.A trading forum is an environment wherein buyersand sellers can fulfill their desire to trade, some-times with small assistance from the forum operator.A trade represents the successful completion ofa negotiation between a willing buyer and seller,meeting in time and place to come to terms on sizeand price.1. A trading forum is a focal point-a cen-tral location-to which trades go, directly orthrough an agent, to initiate and hopefully

complete desired trades. The location may begeographic, as on the floor ofthe NewYork StockExchange or on the "upstairs" block-desk, or itmay exist in cyberspace, as on NASDAQ, theECNs, and most non-US exchanges.2. In addition to meeting in place, the buyer andseller must also meet in time. Even for securitiestrading thousands of times a day, the chance of

Frnsr Quenrrx2004


2/13

\7ewr H.\TecNen

3.

a buyer and a seller of equal size placing simul-taneous market orders is small. The operatorof the trading forum bridges the time/size gap,either by (") encouraging one parry to wait forthe other; usually through a limit order book;or (b) trading with the first parry from his owninventory account, establishing, rhen reversing,the position when the other side arrives.Published price quotes show potenrial buyersand sellers the approximate prices of goods theymay want to exchange. In many markets, thiscan take the form of an obligation on the trad-ing forum operator to trade a specified minimumnumber of shares at the published quotes, thusguaranteeing small but immediate liquidiry forthose who want it.By advertising trading interest to the world viaquotes and sizes available, the trading forumcreates market visibiliry for buyers and sellers.Established rules define for all parties theirrights and obligations while participating innegotiations.Investor confidence in the sysrem results fromfair dealing and timely and accurate reportingand settling of trades.

Frorn the perspective of an individual investor, themarket seems to operare like a vending machine:make your choice, put in your money, and outpops your selection. To all appearances, the pro-cess is totally mechanical. In fact, most exchangeshave standard procedures whereby low-value, low-information, low-risk orders are processed at min-imal operating cost. The market maker operares asa trafEc cop, interceding when necessary to provideinstant liquidiry and move the traffic.Items 2 (time intermediation) and 3 (affirmativeobligation) insert a market maker into the tradingforum. Note the eponymous, self-definitional term:the market maker makes a market by crearing anenvironment wherein buyers and sellers can fulfilltheir desires to trade.

Market making is a simple, low-risk businessto conduct for retail trading by low-informariontraders in small amounrs. Vith the exception oftime intermediation and affirmative obligation, acomputer can easily conduct a trading forum. Thisis why ECNs have been mosr successful automat-ing trading for retail orders, and why conremporaryinstitutional traders seek to tap into the flow byslicing their orders into small, retail-sized pieces.2 Information CenualThe vending machine analogy does not extendto institutional trades. According ro rhe PlexusGroup database of institutional trading, institu-tional orders are very large compared to normaltrading volumes. Over half of the institutional dol-lars traded come from orders that exceed 100,000shares. Eighry percent of institutional dollars tradedderive from orders exceeding a half days volume . Bysize alone these giant orders can easily overwhelmany procedure-oriented trading forum.Thble 1 derived from the Plexus Group databasesof institutional trading, shows that trading vol-ume mushrooms when large institutional investorscommence trading of large positions.Small orders blend into the daily give and take, butthe volume more than doubles the day after a maiorThble I Volumes expand to meer trading demand.Portfolio manager'sorder size divided byprevious days volume

Ratio of volume gainfrom day before orderto day after order

4.

5"

6.

Over 1.0 day! volume0.5-1.0 volume0.25-0.5 volume0.1-0.25 volumeLess than 0.1 volume

222o/o119106100100

JounNar Op INvssrl.rENr MaNecetrcNr Fmsr Quenrpn 2004


3/13

TUE Mer,xrr MAKER rN THE Acr op rsr ECN

institutional trade is begun. The bigger the order,the greater the gain in resulting market volume.\W'eknow how the order initiated, but where did theother side of the trade come from? How did it getthere? These are the ouestions of interest to thispaPer.The classical market model does not apply whensome of the transactors are (a) very large relative toordinary flow and/or (b) imbalanced toward buyingor selling. Simply stated, institutional-sized liquid-ity is seldom there for the taking. Somebody needsto take actions that draw liquidiry to the market.That somebody is a market maker, but not in thesense of normal time intermediation or affirma-tive action. Sitting athwart the information streamsreflecting interest in a stock, a market maker nat-urally accumulates information about past, currentand possible future buying and selling interest: themarket maker is "information-central" about ffad-ing interest in a stock. The market maker suppliesalayer of intelligence on top of the trading forum.Some readers may be surprised at the trading mag-nitudes indicated in Thble 1. Orders of institutionalmagnitude are infrequently seen on the streamingtransaction tape because they seldom meet anothertrade of equivalent size. Thus, these orders are oftenassembled in the market from a large number ofsmaller pieces, and the information-central tradinginteracts with the trading forum activity.Unlike orders in a classical market, these ordersdominate not only because of size but also becausethey represent well-informed opinion. (Thinkabout the effect a Hersheys or a Nestld must haveon the cocoa futures market.) From the marketmakert perspective, massive trade size combineswith informed opinion to make the market makingtasks oftime intervention and affirmative obligationcomplicated and hazardous. An expanded marketmakert role as "information-central" is essential tothe trading of these orders.

These orders require expanded liquidiry to fill theorder. The Random House Unabridged Dictionarydefines liquidiry as "The abiliry or ease with whichassets can be converted into cash." 1We prefer asomewhat broader definition: The abiliry to con-vert assets into cash, or vice-versa when you wantto. at low cost.There are two ways to expand liquidiry but some-thing has to give. It will be necessary to relax eitherthe "when you want it" clause or the "at low cost"portions of the liquidiry definition.One way to think of the market maker's task is interms of liquidiry lying in layers. Strata near the sur-face are easily accessed but thin. Deeper layers canreveal much larger amounts ofliquidiry, but usuallyat higher cost. Figure 1 illustrates the relationshipbetween the size of the trade, the cost of trading,and the strata of liquidiry required to fill orders ofincreasing size. The cheapest, easiest to reach liq-uidiry is at the upper left. As we move down thechart, the arnount of available liquidiry rncreases.\7e will show how it arrives and why it comes at ahigher price.The first step for the trader is to dip into the ebband flow of the everyday liquidiry stream. This is

Size of Trade25% 1O0%

Revealed LiquidiiyFigure I Layers of liquidity.

ctt=Foooo0).9E

Frnsr Quenrrx2004 JounNer Op INr,,rsrvENt trAqNecr,rr.mNt


4/13

W'ayNr, H. WecNsn

"natural liquidiry," where the trading counrerparryarrives as a result of an independent "exogenouslydetermined" decision process. In other words,trading motivation comes from outside-the-box,representing the external decision process.This is not as easy as it sounds. The trader must becareful not ro trade in a noisy fashion that tips offmarket participants such as day traders, technicaltraders, short-term momentum ffaders, and hedgefunds. These traders feed off the flow-informationthat indicates potential trading interest. They knowthat institutions have an information edge and thatinstitutional size will dominate the volume unril iris satisfied. They hope to go along for the ride.But where zi the liquidiry? In todays fractionalizedmarkets, the institutional traders need to search allvenues. Tiaders may try low cost (electronic) venuesfirst, then trading floors. The undone ponion of thetrade often falls into the hands of a block broker toperform the "information-central" role..We do not make the usual distinction betweenbrokers, who represent the trade to the market,and dealers, who trade from their own inven-tory. Both roles are performed at different timesby block dealers, block brokers, and specialists.The important distinction is berwe.r, tr"di.rg forumand informadon-cenrral behavior, not the usualbroker-dealer breakdown.

3 Flow liquidiryFlow liquidiry is like revealed liquidiry except thatit has not arrived yet. The counterpartywill committo trade at some future point, for informational orcash-generated needs. The problem for the institu-tional traders is that they cannot anticipate when,or even whether, someone else will independentlydecide to sell what he wants to purchase. If the orderawaiting liquidiry has a low information conrent,

waiting may not be a problem. In conrrasr, if aninformation "edge" underlies the order, any delaymight be costly if other investors become aware ofthe information. This leaves the trader with a diffi-cult decision: is the best strategy to wait for liquidiryto arrive, or is it more efFective to seek quicker-often more expensive-liquidiry by other tradingtechniques? Since information or trading neces-sity underlies all the largest and most importantinstitutional trades, most institutional traders tilttoward trading fasrer rather than awaiting naturalflowliquidiry.4 Hidden liquidityHidden liquidiry lies one srep away from therevealed and flow liquidiry. The commitment totrade has been made; the order is live. The order maybe known only to the trader holding a large insti-tutional order, or it may be revealed to his blockbroker (e.g. 'participate, do not initiate" orders).In almost all cases hidden liquidiry is larger thanrevealed liquidiry.Quite often the information-central market makeris aware of the liquidiry but trusted to keep it secret.If the market maker senses an opporrunity to puttogether a buyer and a seller, the hidden liquidirycan be revealed to both parties.\X/hy is liquidity hidden ? Consider what would hap-pen if Sheer Genius Invesrment Advisors were roannounce for all to hear "I'm committed to buy50 million shares of AMZN immediately." Theannouncement would attract imitators, some ofwhom would think "Gee, if Sheer Genius is buyingit must be a good srock." Those contemplating sell-ing Amazon would be tempted to hold back, wait-ing to see how high Sheer Genius's inreresr mightpush the stock price. Others, more fleet footed,would think "If I see any liquidiry I can scoop itup now and sell it to Sheer Genius a little later at ahigher price." This behavior is called frontrunning.

JounNer Or INr,'r.srtrENl M-q,NRcevENr Frnsr Quenrpp.2004


5/13

THs Menxrr Maxr,RwrnrAcroprHp ECN

Sheer Genius would lose the share liquidity itwantsto the copycat group. It would ultimately get theliquidity from the committed sellers and the fron-trunner group, but at a higher price. In ail cases, thereactive traders are free-riding on the informationand trading needs of Sheer Genius. Moreover, theiractions move prices ahead of Sheer Genius's abiliryto complete its own trades.Iffree-riding and front-runningwere rare events, wewould not be talking about them here. To the con-trary these are all daylevery day occurrences. Theyare, for example, the raison d'ene of the day traders.They are a major concern to institutional traders.As a result, large trading interest is seldom shownpublicly in order to protect the proprietary value ofthe order. Information on the desire to trade willnot be revealed without a suitable quid pro quoi theinformation will be traded either for an executionor for information-trading or fundamentd-ofperceived equd value. It will not be intentionallyreleased to untrustworthypartieswho can gain fromthe knowledge at the expense of the order initiator.A primary dury of an institutional trader is to pro-tect the value of this information to the advantageof his employert clients.Note that the initiating institution acts from aninformation edge, particularly with the largestorders. Tiading is necessary to capture the infor-mation advantage. But the initiator cannot act inthe marketplace in size unless someone is willingto trade with him. The order cannot be completelyhidden; it must be known to someone who (1) canbe trusted to protect the information and (2) is sit-uated to know when a counterparty arrives. if bothbuyer and seller notify a market maker of potentialtrading interest, a.k.a. hidden liquidiry the marketmaker can effect a trade.The importance of hidden liquidiry is emphasizedby the Liquidnet and reserve book techniques of

gathering liquidiry while revealing it only when itleads to a trade.Once the supplies of non-hazardous immediateliquidiry are exhausted, the institutional trader facesone of two choices: continue to wait for liquidityflow to occur naturallv or attempt to draw out theother side.

5 For-hire liquidity'We described above trades occurring through thematching ofexogenous trading decisions. The inter-esting question is what happens when no "natural"arrives to balance buying and selling interests. TheIast two categories of liquidiry derive from marketparticipants who respond to calls for liquidiry. Theywill accommodate liquidiry demand by buying at adiscount or selling at a premium. In other words,they sell liquidiry to any party anxious enough topay for it. The information edge these traders workfrom is endogenously determined, inside-the-boxinformation about the market. The order itself isthe inside information. In other words, the initia-tor is not only selling stock, it is selling part of itsinformation edge in order to be able to trade.From the institutiont perspective, liquidityproviders can be incented by giving up a portion ofthe potential alpha in order to coax out the desiredliquidiry. The trade will be profitable as long as thepayment for liquidiry does not exceed the potentialalpha. By our experience, portfolio managers aremore attuned to buying liquidiry with alpha thanare institutional traders. This leads to the anoma-lous conclusion that costs can be too low if they leadto lower rather than higher portfolio return.There are many sources of endogenously deter-mined liquidiry. (1) A market maker might under-take this role. (2) The block trading desk ofa large brokerage firm might use its capital to

Frnsr Quanrsn 2004 JouwRr Op INvesrrrENr MeNecr.rr{Br'rt


6/13

\X/evNn H. WecNan

accommodate better customers. (3) Large brokersalso staffproprietary trading desks. They are stand-alone profit centers; if they sense a profitable tradingopportuniry they will seize it. (4) Alternatively,aggressive non-broker traders such as hedge fundsmay take a short-term position if they believe theycan profit. One of the roles of information-centralis to alert these potential liquidiry sources.It is not difficult to distinguish berween frontrun-ners and liquidiry providers. Suppose the word goesout that Sheer Genius wanrs to buy stock. Thosewho sell as a result of this information are liq-uidiry providers; those who buy as a result of thisinformation are frontrunners. Institutional tradersdevelop ulcers when calls for liquidiry draw morefrontrunners than liquidiry providers. They suspectthat market makers play a nefarious role. They areparticularly concerned about holes in the invisible"Chinese'Wall" between the block traders and theproprietary traders.At first glance, signaling these endogenous liquidiryproviders appears to breach the confidential natureof the order. \fhile knowledge of the trade is avaluable proprietary asset of the initiator, it is alsoknowledge that the market makers purvey as theirstock in trade.The market makers aid their customer by trans-mitting information that attracts liquidiry. Thedistinction is not clear-cut: it is impossible to drawa black and white distinction berween seeking liq-uidiry and violating confidentialiry. No bright linedistinction exists berween "trying to find the otherside" and "tipping off his pds to frontrunningpossibilities." The market maker cannor accelerareliquidiry arrival without revealing trading interest.Nor is the institutional trader able to completehis/her order expeditiouslywithout revealing at leasta peek. The trader must balance the costs of reveal-ing information against the cost of failing to findliquidiw.

This situation frustrates institutional traders. Theymust trust but cannot verify. Potential conficts ofinterest are rife among the parties they must dealthrough. Even worse, they have no abiliry to mon-itor the market makert information dispersal. Nopaper trail exists, no phone records can be relied on;information could be passed on by a code, a blink,a hesitation, or silence at the pregnant moment.It gets worse. fu information moves down the chainof trading, other parties with no connection or obli-gation to the customer tune in. A weak marketmaker may have neither the capital to finance thetrading nor the treasured relationship with the insti-tution, and may "daisy chain" the trading by layingoffpositions to other brokers or hedge funds. Thisarrangement clearly exposes latent trading demand.No wonder institutional traders would orefer tofind the other sid,e without the use of a m"rk.tmaker!Liquidiry providers can be drawn from a myriad ofdirections. The most innocent way to find poten-tial sellers is to call up those who are known toown the security and those who have been recentsellers. This is clearly part of the information-central role. Another source would be someonewho could arbitrage between a sale of, say SonyADRs, while simultaneously buying Sony sharesin Tokyo. An options market maker might find away to profit from an arbitrage against the options.A hedge fund might sell stock and buy a mathe-matical hedge of other companies to control theexposure.If these are such efFective methods of raising liq-uidiry why did the Sheer Genius traders nor ffythem themselves? They will not wanr to call otherinvestors because they would just as soon rhe othersdid not knowwhat theywere doing. As the old say-ing goes, "Does M".yr tell Gimbels?" The traderhas neither the time, the resources nor, particularly,the contacts.

JounNm O r It vEsrvrNr MeNecErr,reNr Frnsr Quanrrn 2004


7/13

THs Mamer Merr,RrNrun Acsorrss ECN 10

Ah, but the market makers dohave the contacts. Itis the heart of their business. Maintaining this net-work of contacts is time consuming and expensive.Only someone with an on-going, every day missionto be privy to everFhing knowable about a stock'ssupply and demand can profit enough to cover thehigh network maintenance cost.'W'e have one more category of liquidiry to tend to;then, we will return to the issues involved in findingfor-hire liquidiry.

6 Last-resort liquidirySuppose a company unexpectedly restates earnings,leading to a panic of intensely motivated sellers.Everyone, including the for-hire liquidiry providers,hangs back until price reaches a bottom. In thissituation, there are no buyers at the current price,and stock for saie will overhang the marker untilthe price reaches a level where last-resort liquidirycomes into play.'Who are these buyers of last resort? The junkmen;the deep contrarians, who compare price to assetvalue and are motivated to trade by a deep, deepdiscount. Donald Keiml has documented how onefund manager, Dimensional Fund Advisors (DFA),systematically adds value by standing as buyer ofTable 2 Levels of liquidiry.

last resort. The lower line in Figure 2, taken fromwww.dfafunds.com, shows that stock prices declineon average 10olo over 20 days prior to DFA blockpurchases, then bounce back and stabilize. This isthe price pattern we would expect ro see from abuyer oflast resort.Value investors are constantly on the prowl for thesesituations and will likely uncover them withoutmarket maker assistance. They will step in whenthe price falls farther than the fundamentals. Still,these are rislqy trades that demand big price conces-sions to offset the risks of buying too soon or buyingfatally damaged goods.TabIe 2 summarizes the above discussion of thelevels of liquidity.

slRwlul*vTiaSs*

fi,frVLt*kTia4*$

fiS!-s frl?il* irr T'ade *afp ff*iFigure 2 "Last resort" liquidiry pattern.

!,1,\.*

,8.. r,:a*w.a'fJ a:::3tt::ilf.t

fiuys S*ty

Vwffiz)*

Level of transaction Nature of liquidiw orovider Type of liquidity Role of market makerRevealedFlowHiddenFor-hireLast-resort

Publicly pre-committed\X/ill decide to tradein a short timeframePrivately pre-commimedStand-by commitment at

a priceValue investor

NaturalNaturalNaturalSituational(liquidiry seller)Natural (contrarian)

Ti'affic copThaffic copInformation-central

secret keeperInformation-central

bush beaterTiaffic cop

Frnsr QuenrE.n2004 JounNer- Op Ixr,'p.srN,reur MeNecEnmNr


8/13

11 WeyNE H. WAGNER

7 How information moves stock pricesSometimes liquidiry is plentiful, sometimes it isscarce. It becomes plentiful when there are rea-sons to trade; i.e. news. News causes the price tomove about as different players react. Some willfeel the news portends higher prices and want tobuy; others chime in with sell orders as they feel theprice becomes over-extended. As the price moves, ittrips the triggers-both price andvolume-of othertraders who were waiting to buy or sell.Tiaders sayliquidirybegets liquidiry. All this activirywill generate volume leaders, financial news cover-age, Maria Bartiromo headshots, and other levels ofexcitement. Once news-based traders create liquid-iry momentum traders, hedge funds, and othersstep in to try to capture a profit from temporarysupply/demand imbalances and over/understatedprices. Technical traders study trading patterns fortelltale footprints and jump in based on perceivedsupply/demand balance. The liquidity generated bythis activity may be twice to ten or more timesnormal volume.Figure 3 from bigcharts.com is typical; it couldhave been selected at random. During the last weekof July, 2001 Cooper Industries traded a couplehundred shares per day. A merger announcementon August 1 caused the price and volume spike.Note how the volume spikes correspond to larger

Figure 3 Volume jumps when price jumps.

changes in price, and note how rapidly the activirydies down in between.Once the news efFect passes, activiry setdes down.\fith no market, industry or company reason totrade, the market sees only liquidiry traders. Itbecomes difficult to trade large volumes withoutsparking substantial price efi[ects.'We describe below the process through which fun-damental information moves market prices, andhow news affects stock price, played through thescrambling necessary as buyers and sellers seek theliquidiry to optimize their strategies.1. Low-volume liquidiry trading prevails untilinformation arrives. This is the classical mar-

ket of no-news situations; only liquidiry tradersactive, Iow volume and wandering, trendlessprice.2. Outside-the-box information sparks tradinginterest among naturals, leading to tradinginterest and price moves.3. Their activiry leaves readable footprints thatattract inside-the-box copycats who accelerateprice movement. Volume increases.4. Price movement and signs of liquidiry imbal-ance draw out for-hire accommodators. Vol-ume explodes with the rapid buying and sellingof the liquidiry providers.

7.

Ultimately, price moves to the point whereinformation-driven naturals drop out, copy-cats bail, and liquidiry accommodators flattenpositions in a game of musical chairs.The situation returns to low volume liquidirytrading as the last resort traders come in andmop up what is left.All the while, the market maker coordinatesthis frenzied pas de troupe.

This is a new domain in economics called marketprocess theory which focuses not on the efficiency

5

6.

JounNal Or INvEsrraENt MaNecErvrp,ur Frnsr Quenrrl'2004


9/13

11 \TnvNE H.'WecNp,n

7 How information moves stock pricesSometimes liquidity is plentiful, sometimes it isscarce. It becomes plentiful when there are rea-sons to trade; i.e. news. News causes the price tomove about as different players react. Some willfeel the news portends higher prices and want tobuy; others chime in with sell orders as they feel theprice becomes over-extended. As the price moves, ittrips the triggers-both price andvolume-of othertraders who were waiting to buy or sell.Thaders say liquidity begets liquidiry. All this activirywill generate volume leaders, financial news cover-age, Maria Bartiromo headshots, and other levels ofexcitement. Once news-based traders create liquid-ity, momentum traders, hedge funds, and othersstep in to try to capture a profit from temporarysupply/demand imbalances and over/understatedprices. Technical traders study trading patterns fortelltale footprints and jump in based on perceivedsupply/demand balance. The liquidiry generated bythis activiry may be twice to ten or more timesnormal volume.Figure 3 from bigcharts.com is rypical; it couldhave been selected at random. During the last weekof July, 2001 Cooper Industries traded a couplehundred shares per day. A merger announcementon August 1 caused the price and volume spike.Note how the volume spikes correspond to larger

LW1&sss454qss@,o 'to;o

changes in price, and note how rapidly the activitydies down in between.Once the news effbct passes, activity setdes down.\V'ith no market, industry or company reason touade, the market sees only liquidiry traders. Itbecomes difficult to trade large volumes withoutsparking substantial price efiFects.W'e describe below the process through which fun-damental information moves market prices, andhow news affects stock price, played through thescrambling necessary as buyers and sellers seek theliquidiry to optimize their strategies.

1. Low-volume liquidiry trading prevails untilinformation arrives. This is the classical mar-ket of no-news situations; only liquidity tradersactive, low volume and wandering, trendlessprice.2. Outside-the-box information sparks tradinginterest among naturals, leading to tradinginterest and price moves.3. Their activiry leaves readable footprints thatattract inside-the-box copycats who accelerateprice movement. Volume increases.4. Price movement and signs of liquidiry imbal-ance draw out for-hire accommodators. Vol-ume explodes with the rapid buying and sellingof the liquidiry providers.5. Ultimately, price moves to the point whereinformation-driven naturals drop out, copy-cats bail, and liquidity accommodators flattenpositions in a game of musical chairs.5. The situation returns to low volume liquidirytrading as the last resort traders come in andmop up what is left.7. Nl the while. the market maker coordinatesthis frenzied pas de troape.

This is a new domain in economics cailed marketprocess theory which focuses not on the efficiency

tgit4iie. awi

Seo oct H6u 16. n2Figure 3 Volume jumps when price jumps.TouRNel Or INvEsrHarNr Meuecrnrxt Frnsr QuerrEn 2004


10/13

Tss Melxsr Merrn w rur Acs orrHr, ECN T2

of the market mechanism but the acrions whichtake place in the market arena. The informationand communication behavior of the participants iscrucial in this analysis. The market is seen to be adiscovery/learning process which rewards the bestinformed participants with the greatest knowledgeprofits.2The market for individual stocks stays in theno-news liquidiry condition most of the time,yet most market activity occurs in intense news-adjustment periods. In these periods, the marketmaker plays the key information-centrai role.

8 Nash equilibriumIn 'A Beautiful Mind," Sylvia Nasar3 describescooperative games as conficts in which players havethe opportunity to communicate and collaborateand are able "to discuss the situation and agree ona rational joint plan of action, an agreement that isassumed to be enforceable." In cooperative games,players form coalitions and reach agreements. Thekey assumption is that there is an umpire around toenforce the agreement.The umpire, the market maker, is indispensable tofinding this Nash equilibrium. The market makerknows what has happened, who has shown interest,who might be willing 16 ffads-xnd who might bestanding by to provide for-hire liquidiry and drawthem into the market.It is common to speak of the stock market as a rwo-person, zero-sum game, one in which one person'sgains are the other person's losses. However, as Nasarpoints out, two-person, zero-sum games occur inboard games, but are extremely rare in the world ofpolitics and economics: "Pure conflict, in which theinterests of two antagonists are completely opposed,is a special case; it would arise in a war of competeextermination, otherwise not even in war." This is

hardly a description of the day-to-day operation ofsecurities markets.\Vhile every securities transaction occurs berweena single buyer and a single seller, their interests arenot in direct conflict. Rather, each has agreed to theterms because each defines their objective in termsof their needs, their utiliry. \7hi1e neither achievestheir pure, unfettered maximization of their or theirinvestors' utiliry each feels that the transaction is thebest that could be done under the circumsrances.The investors who have done the research realizethat the advantage of uading a zero-sum man-ner will be limited in its economic advantage bythe paucity of "flow liquidiry." In order to fur-ther exploit the information advantage, They mustforego a portion ofthat research advantage ro traderswho will trade only if they can piggyback on thatresearch advantage.This is an everyday, all-the-time occurrence. Liq-uidiry is not a free good. Rather, it is a cost of beingable to implement good investment ideas in institu-tional size. Those who satisfy the liquidity demandreceive compensation for extending the privilege.So this is a Nash equilibrium. No player is zero-sum oriented toward winning at all costs. Rather,the manifest solution is to encourage players willingto provide liquidity for trading profits so that thosewho develop the research insights can secure at leastpart of the profit.

9 'What ECNs do wellAn ECNt first advantage is the computerizationof manual efforts. Expensive labor is replaced bydirt-cheap computing power coupled to exrensiveelectronic connectivity. Response time shrinks anderror rates tumble. These features guarantee the suc-cess of electronic trading in the retail market. This

Fnsr Querrep.2004 JounNel Op INvEsrrmNt MeNecEnrExr


11/13

L3 \VavNr H. \TecNEn

is not pleasing to those whose skill, experience, andwealth are tied to floor trading.Automated price discovery can be another advan-tage of an ECN. One of the market maker'straditional functions is to find the price point thatsatisfies the greatest trading interest. This functionis performed simply in a crossing network and moreelaborately in the defunct Optimark and AZX sys-tems. Even here, trading occurs in the light cast bythe prices in the primary market. 'W'ithout the bal-last of central market quores, ECN prices quicklylose touch with realiry.Tiaders love the abiliry of an ECN to main-tain anonymiry while performing some of theinformation-central role. Institutional traders pre-fer anonymous trading because they fear copy catsand frontrunners who observe their actions verycarefully. Because of this suspicion, alternativetrading mechanisms such as third market brokers,Instinet, Optimark, and Liquidnet generare initialenthusiasm and in some cases significant marketshare by offering to take the market maker out ofthe process.Some ECNs have tried to exclude less desirabletrading partners. For example, Instinet, an abbre-viation of Institutional Network, was designedexclusively for institution to institution trad-ing. Instinet failed to generate a profit as longas it self-limited its clientele to institutions. Itgained popularity and profitability only when bro-kers were allowed to interact. The advantage of"cutting out the middleman" failed to overcomethe added search costs imposed by an inabil-ity to find trading partners outside the exclusiveclub.The newer Liquidnet system has a similarinstitutions-only restriction, but so far appears tobe avoiding the frustrations Instinet experienced inits early years.

10 '\il7hat ECNs cannot do wellECNs are effective at reducing search costs andaccessing natural liquidiry. They do not seem asproficient at setting prices, operating information-central, or assessing risk.10.1 Setting pricesDirectly or indirecdy, ECNs set prices by piggy-backing off the centralized market. There is muchmore to finding a price than simply observing thelimit order book or recent order flow. Throughexperience and a network of contacts, the marketmaker develops a sense of what might lie aroundthe corner. It is beyond our current technologyto create a computer algorithm that receives, pro-cesses, and weighs this intense yet nuanced strearnof information.Yer, somebody has to set the price : this would seemto intimate that the market maker knows the price,but on closer inspection this turns out ro be untrue.The market makers set a plausible price, keep theirexposure low, and make prompt corrections if theyturn out to be wrong. This fine tuning keeps themarket makert exposure under control. A marketmaker makes money as a highly informed and veryagile facilitator, not as an investor.In todays markets the stream of trade prices andquotes is broadly available at low cost. A battleis brewing over whether the exchange can claimproperty rights on the creation and maintenanceof the price discovery process. Price discovery isa major product of an exchange, although we sel-dom price its economicvalue. ECNs and secondaryexchanges that free-ride on the dominant markettprice discovery process can offer transaction servicesof similar qualiry without having to bear the costsof price setting. The question ofwho owns properryrights on price discoverywould seem to raise publicpolicy questions worthy of considerable debate.

JounNer Op INvrsrnrsNt MeNecEivlENr Fnsr QuenrEp.2004


12/13

THs Mamnr MarBR tN rirr, Acr or rns ECN t4

10.2 Information-centralfu information-central, the market maker is adeptat accessing the lower strata of liquidiry. Despite theadvance of automated markets, the direct liquidiryseeking price inquiry process remains where it hasbeen always-in the hands of the market maker.In no-news conditions, the information-central roleis unimportant and ECNs perform well. Thingsbecome a lot more interesting when supply anddemand are out ofbalance. The institutional traderswho cannot afford to wait need to draw out a liq-uidiry when there is no buyer on the ECN for thestock they want to sell. The ECN can do nothingwith non-matching orders other than sit on them orreject them back to the originator.'We suppose rharan ECN could be programmed to send messagesto recent trading interest; but, without human dis-cretion in the process, it would quickly deteriorateinto a deluge of spam.

10.3 Risk assessmenrSelling a security transfers the risk of owning ro rhebuyer. \(hen stock is sold to a for-hire liquidiryprovider it remains actively for sale until he canreverse the position. Unlike the other strata of liq-uidiry for-hire ffaders, like market makers, are notinvestors.A buyer operating a temporary way station on theroute to the ultimate buyer bears risks proportionalto the exposure in dollars and holding time.In contrast to the anonymiry ofscreen trading, floortrading is conducted face-to-face and, rherefore,not in an anonymous manner. Experienced floortraders lament the lack ofcontextual information inanonymous screen markets. \Vhen orders have nodiscernable background, insight into the markettreaction is hardly possible. Nor is there any loyalty

or social sanction possible to discipline wrong-doers. Thus, opacity adds risk for all participants,especially market makers.

The risk is ever-present in market making and thesupply of for-hire liquidiry. In the author's opin-ion, this risk is intuitive and emotional; somethingfelt in the pit of your stomach rather than calcu-lated with mathematical precision. Risk assessmenris essential to the workings of a market. 'Withoutconsequences, we face a situation of moral hazard.Like the dead man throttle on a locomotive, thehuman perception of risk is a necessary governor roprevent a runaway train.MadoffSecurities is an example in point; the com-puter makes markets until pre-specified volumebounds are reached; e.g. our inventory of AMZNstock is too large; we need to lay off the risk. Eventhough the routine processing is done by computer,the boundaries are set by humans, and rhe actionstaken to control the risk require human artenrion.Indeed, it is under extreme market conditions thatelectronic markets show the most strain. Theyare designed for foreseeable situations; when theunforeseeable happens they have no intuition orexperience to fall back on.Except in science fiction, it is impossible to inflictfinancial or emotional pain on an unfeeling com-puter. 'We do not know how to teach a computerto react in the same manner that a human beingwould. There is a good chance we never will.

11 ConclusionSchumpeter spoke of innovation as reatiue destruc-tion, and it abounds in the securities markets. Inno-vation and technology are changing the market andwill continue to do so. The comperition betweennew and traditional market places is healthy, forcing

Frnsr Quenrer.2004 TounNel Or INvr,snr,rENr MeNecrir.mNt


13/13

r5 WevNr H. \(.tcNnr.

all to hone their unique advantages. Technologydecreases the importance of traditional high coststructures and challenges the franchise power ofbrokers and exchanges. Automation increases theffansacdon speed, creating new opportunities formanagers with hair triggers. Reducing costs makemore investment ideas actionable, which increasesliquidiry. Increased liquidiry leads to the ability tomanage larger pools of assets. At the bottom line,this enhances investor Power.\7e come to the conclusion that the computerwill not soon obsolete the traditional functions ofthe human market maker' The information-centralrole is too essential to the process, and requiresknowledge, skill, continuing relationships, and riskassessments that are now impossible to program intoa comPuter.Yes, there zi an indirect cost: in order to find theneeded liquidiry, large institutional uaders mustgive away an informational advantage to inter-mediaries. Fairness and integrity require that theinformation is not used to the harm of the orig-inator. For powerfrrl institutions, the integriryof the system leans on a continuing relationshipwith important monetary implications at stake'Monitoring and feedback play a critical role'In 1989 the author attended a conference on thestate of the marketwhere the head of a major sellsidedesk warned: If brokers were not adequately com-pensated for services offered, they would becomei.r, " partner to the buyside and more of anadversary. Prophetic? The old salts in the tradercommuniry believe this has already happened' Theyfear that the institutions' slipping influence overmarket maker loyalry as investors such as hedgefunds whose high turnover and higher commissionscrowd out the interests of institutional investors'

fu Picot et al. conclude, "Complete replacement ofhuman intermediaries will not take place. But it islikely that there will be an unbundling of the activ-ities of intermediaries so that consuhatiue femphasismine] and pure order handling activities are sepa-rated. Standard transactions will become more andmore automated, even their price discovery as longas the orders are atomistic and one order has nosignificant impact on the price. Intricate transac-tions will still need human intermediaries but maybe conducted with electronic support."W'e are still searching for the newNash Equilibrium'The future promises to be interesting; youwill wantto stay tuned.

AcknowledgmentsProfound thanks to Mark Edwards, Larry Harris,and Gary Gastineau for helping me turn a bowl ofconfusion into a more coherent Paper.

Notes1 Keim, Donald (1999). 'An Analysis Of Mutual Fund

Design. The Case For Investing In Small-Cap Stocks'"Journal of Financial Economics 51.2 Picot, Arnold, Chrisdne Bortenlaenger, and HeinerRoehrl. The Automation of Capitai Markets' Linktowww.ascusc.org/jcmc/vol1/issue3/picot'html' This one-of-a-kind paper is an undiscovered jewel for anyone whowants to think straight about market structure'3 N"sar, Sylvia (1998) . A Beautiful Mind: The Life of Math'ematical Genius and Nobel Laureate John Nash' New York:Touchston Books.

Keltwords: Market maker; ECN; market liquidity

JounNer Op INvEsrvrNr MeNecetr'tENr Frnsr QuearEn 2004

Documents

The Marketmaker in the Age of the ECN