MATIF, the Paris-based derivatives exchange, no longer exists. It opened for business as a traditional floor-based exchange on February 20, 1986. Thirteen years later, in 1999, it was absorbed into a group of French exchanges under the name Paris Bourse. A year later the Paris Bourse was itself absorbed into the pan-European exchange called Euronext NV. ¹ During its last year as an independent exchange, MATIF became 100% screen-based.

Unlike its Chicago cousins, which have spent the better part of two decades making the shift, MATIF's transition from floor to screen took just over six weeks. This scared the old floor-based markets to death. But the rapidity of the transition was not just a matter of market participants voting with their fingers. It was helped along by the peculiar French habit of addressing all problems with a good, old-fashioned strike.

This is what happened: MATIF announced that on April 3, 1998, the floor and new electronic system would run side by side for the interest rate contracts, which dominated business activity. Fees would be the same on both platforms, giving neither one a fee advantage. Participants would have a choice as to whether they placed orders on the floor or in the computer. It was a hybrid system, and it was up to the market how fast the transition from floor to screen would be.

The first strike occurred on February 2, 1998, about a month before the screen system was due to go live. Locals walked off their jobs for four days to protest the requirements they had to meet to participate on the screen. Volumes fell dramatically, and the strike worked because MATIF cut in half the trading requirement locals needed to meet to get free access to the system. Initially the requirement was 250 contracts per day, but this number was renegotiated down to 125 contracts a day.

Then, in early March, the majority of MATIF's locals went out on a second strike, deciding that they wanted even better terms than they had been granted after the first strike. They also wanted lower fees, even though fees had already been cut in January. They were also upset that a special $8 million fund had been set aside to help brokers make the transition to screens, but nothing similar was done for locals. Again, the strike paid off. The exchange offered, for six months, to waive all trading fees, to provide screens at no cost, and to create a $1.6 million bonus pool for locals to share based on the number of contracts each traded. Despite the offer, many locals remained on strike, demanding tens of millions of dollars more. ²

Because of a hardware problem, there was a five-day delay, and the side-by-side trading of interest rate futures did not begin till April 8. The striking floor locals, through their absence, made it difficult to get trades executed on the floor and made the screen even more attractive than it would have been had the floor been fully open and accessible. By early May, 90% of all interest rate trading was on screen. CAC-40 futures went straight to screen on April 8, and the single stock and CAC-40 options had moved to screen some time before that. In less than seven weeks the market had clearly spoken, and the floor was closed on June 2, 1998. MATIF was now 100% electronic and became the 13^th derivatives exchange to become so (see Table 3.1).

There are always glitches in startups, and MATIF had its share, but the exchange got undeservedly bad press for a few weeks that raised concerns on both sides of the Atlantic. On July 23, 1998, the French bond market tumbled when, during a five-minute period, there was a strange series of market sell orders for a total of 10,607 contracts. The Notional bond dropped 149 basis points and then snapped back. All the market orders came from the Salomon Smith Barney office in London. A few trades (for 119 contracts) were busted; most were not. ³

Salomon Smith Barney was one of MATIF's largest clearing firms, so when the firm claimed this activity was due to a bug in MATIF's NSC software, MATIF hired two independent auditing firms to investigate. This was also a sensitive issue because this same NSC software was the foundation for the new, improved version of CME's GLOBEX system.

Even before the investigation was complete, MATIF implemented three changes to prevent a repeat performance. First, it eliminated the “quick trade” button that allowed traders to trade by pressing a single button. Now a trade would take two buttons pressed simultaneously. Second, it created a temporary freeze feature that would halt the market for two minutes if the bond's price ever fell 25% or more. Third, if a very large, unusual market order came into the system, the market would be halted till the exchange could verify it from the trader. This typically took much less than a minute.

When the two independent auditors, CAP Gemini and Kroll Associates, handed over their reports, it was clear that the problem was not in the NSC system but rather had occurred when the London trader inadvertently leaned against the keyboard, sending a flurry of orders. ⁴

In mid-1997, LIFFE was sitting on top of the world. It had just moved up to become the second largest exchange globally, after the Chicago Board of Trade (CBOT) and before the Chicago Mercantile Exchange (CME). It was a teenager, only 15 years old, and had just surpassed the 99-year-old CME, on which it had modeled itself. It had a great product line, the star of which was the German bund. It was also a bit more aggressive about electronic trading than were the U.S. exchanges. It had been the first exchange in the world to create a screen-based after-hours trading system.

However, LIFFE had not been aggressive enough. As mentioned earlier in this book, Frankfurt-based DTB was doing all the right things to ensure that it was able to bring the bund to Germany. It had effectively placed DTB terminals at practically no cost in the United States, France, and right in the heart of London. As a clear sign of what LIFFE was up against, in late March 1998 one of LIFFE's most prominent members dramatically resigned from the LIFFE board. This board member, David Kyte, one of the biggest floor traders in the bund, said he had no confidence in either LIFFE's chairman or its chief executive. He felt that intense competition the exchange was getting from DTB should cause it to immediately throw the bund onto automated pit trading (APT), the electronic system LIFFE had been using for after-hours trading. There wasn't, Kyte felt, enough breathing room to wait and gradually migrate products over to LIFFE Connect over the next 18 months. DTB already had a greater market share than LIFFE and was breathing down LIFFE's neck.

Though he was spot-on in his forecast, Kyte's own business behavior helped DTB prevail. At the time of his resignation, he had already obtained 50 DTB terminals to allow half the hundred traders in his firm to trade the bund in Frankfurt. If even LIFFE board members were abandoning ship, what chance did LIFFE have?⁵ Within four months of Kyte's dramatic gesture, both the chairman and the CEO had been sacked or had left the organization.

Despite LIFFE having one of the first after-hours trading systems and despite developing a world-class trading platform in LIFFE Connect, Kyte was right: LIFFE had started too late and got blindsided by DTB. Within nine months of Kyte's resignation from the LIFFE board, DTB had captured 100% of the business in the German bund, LIFFE's flagship product for almost a decade. Three years later, in January 2002, a diminished LIFFE was absorbed by pan-European market Euronext (formed by the merger of the Amsterdam Stock Exchange, Paris Bourse, and the Brussels Stock Exchange).

LIFFE's technology platform, LIFFE Connect, did provide a smooth migration from pit trading to electronic trading. All the markets under Euronext began trading on the LIFFE Connect platform. In addition, the merged Euronext.liffe now sells its technology globally. Today, a number of exchanges utilize LIFFE Connect, such as the Tokyo International Financial Futures Exchange (TIFFE) and, until its 2007 merger with CME, the CBOT.

The United States is no technological dwarf. Why, then, did it take so long for the country to get into the screen-based trading game? The answer is simple: U.S. exchanges were owned by their members, by the men and (much less often) women who traded on the exchange floors. The members were making a good living in the old floor system. Why should they throw all this away for a new system that wasn't yet proven and, more importantly, on which they might not be able to compete effectively against a younger generation raised on videogames? The exchanges that would act as pioneers were those without these legacy problems—the new ones, the ones with little to lose.

But the big, member-owned exchanges did put a toe in the electronic waters by adopting limited electronic trading. Specifically, several U.S. and European exchanges, along with the Sydney Futures Exchange, created electronic systems that would be available only after regular trading hours, to solve a different problem and, most important, in a fashion that would not compete directly with their own floors.

This different problem was the need to prevent offshore competition and extend their product monopolies into other time zones as countries in those other zones began to set up their own exchanges. It was an attempt to capture for a given exchange as much of the 24-hour day as possible, or at least the portion of the day in which potential customers were at work. Initially these new exchanges in other time zones would list futures contracts based on their own local equity indexes, their own domestic government debt, and their own commodity products, but there was the chance that they could begin to list U.S. Treasury bonds or Eurodollars or currencies or internationally traded grains or petroleum products and thus compete directly with the big, established U.S. and European exchanges. Even though Japan and other Asian countries were experiencing a financial awakening and were beginning to trade in a big way on the CME, CBOT, LIFFE, and MATIF, these American and European exchanges needed to make their markets accessible during Asian business hours if they expected to maintain this business.

There were at least three ways to extend trading hours and ensure that traders living on the other side of the world would keep trading in Chicago, London, and Paris. The most obvious solution was to simply extend pit trading hours by bringing on a new set of traders to make markets throughout the U.S. or European nights. It was an obvious solution because the physical infrastructure was already there. Just as factories are kept going 24 hours by bringing in three shifts of workers, why not do the same with exchanges? The CBOT actually tried a limited version of this approach with a three-hour evening session for interest rate futures and options in 1987, but it was not successful. The New York Cotton Exchange's (NYCE's) FINEX Division, created for financial instruments, added an evening floor session in 1992, which, amazingly, continued through mid-2007. The problem with night shifts was that not only did the exchange have to attract a new set of fresh traders to the floor, the brokerage firms needed to staff the trading floor booths as well as all other aspects of the trading process. It was a labor-intensive, expensive solution that was ultimately not efficient.

A second approach was to extend trading hours by setting up a trading floor in the targeted time zone. The CME did something close to this in 1984, the same year INTEX set up the world's first fully electronic exchange. It did not actually open its own subsidiary overseas, but it became closely involved with setting up a new exchange in Singapore. Specifically, it established a link with the new Singapore International Monetary Exchange (SIMEX) that tied one of its most important products to an identical product being listed at SIMEX and led to a growth in trading at both exchanges. The product consisted of three-month Eurodollar deposits and the deal, called the Mutual Offset System (or MOS), allowed the CME Eurodollar contract to be traded on the SIMEX floor and then, if the trader wished, his position would be moved to the CME clearinghouse in Chicago and offset on the CME trading floor. In fact, the CME was so committed to establishing a base in the Asian time zone that it actually aided the Monetary Authority of Singapore (MAS) in setting up SIMEX by sending over both staff and traders to train the SIMEX crew.

A more direct attempt to establish a floor in a different time zone was NYCE's attempt to set up a trading floor in Dublin for its FINEX Division's currencies in 1994, which did create a modest amount of business for a number of years. The most recent move in this direction was NYMEX's 2006 creation of a floor-based subsidiary, first in Dublin and later in London, for its energy products. NYMEX's purpose was a little different from NYCE's in that it was attempting to poach some of the experienced traders from a competitor that was replacing its floor with screens. The competitor was the International Petroleum Exchange (IPE), which had been taken over by another closer-to-home competitor, the Atlanta-based Intercontinental Exchange (ICE). Both of NYMEX's attempts to create new European floors failed.

This brings us to the third approach to extending the trading day to protect existing monopolies, and that third approach is after-hours electronic trading. The idea here was to create and use screen-based trading, but only after regular pit trading was done for the day. This avoided any direct screen-to-floor competition and minimized pushback from exchange members currently making their living on the trading floor. So, CME, CBOT and LIFFE products would be available for trading, not only during their own business days, but also during the business days in other time zones. It is worth taking a closer look at some of these after-hours systems.

For almost two decades, the Chicago exchanges, along with NYMEX in New York, walked the difficult path of trying to adapt to the modern world but at the same time protect the livelihoods of their floor-bound members. In other words, the exchanges continued to add technology, but only on the periphery of trading. There were technological improvements in getting orders to the floor, clearing matched trades, and bringing fundamental information and analytical systems to the trading floor. But the execution of trades continued to be done by human beings yelling at one another. That ultimate right of exchange members to execute orders, make markets, and trade against orders, and to do so via hand signals and voice, continued to be preserved as much as possible. Even as electronic trading began to make its way out of the after-hours closet and into regular trading hours, the costs of electronic trades were kept high or speed bumps were constructed to prevent electronic trading from looking too attractive.

But everything changed in the summer of 2003. Eurex, then the biggest exchange in the world, announced that it had plans to open a subsidiary exchange in Chicago, and it began to seek regulatory approval to do so. The Chicago exchanges were scared to death. The smart money was on Eurex. Everyone remembered the battle of the bund, the battle to the death between Eurex¹³ and the LIFFE. The London-based LIFFE lost the German government bond to the quick, cheap, transparent, screen-based Eurex. And everyone thought that this would be repeated as screen-based Eurex came into Chicago and took on the floor-based CBOT head to head. Now, it's true that the CBOT had its own after-hours electronic system that it had even started to expand to side-by-side simultaneous floor and screen trading, but the overwhelming bulk of trading on the CBOT was still floor-based. In addition, even though the initial attack was to be on the CBOT products—that is, the Treasury futures and options—the CME felt no less threatened. In fact, its fear was well founded, since after an initial sustained attack on CBOT Treasuries, Eurex U.S. did turn its attention to CME currencies.

So together, the Chicago exchanges launched a counterattack. The first thing they did was to throw up every regulatory barrier they could. For example, they ensured that Congressional hearings were held, given the fact that this foreign board of trade, this German exchange, intended to trade in futures and options based on the sacred and precious government securities of the United States. They also did everything they could to slow down the approval process at the Commodity Futures Trading Commission by questioning every single aspect of the application made by Eurex for the new exchange. Second, since what Eurex had going for it was a transparent, electronic, relatively inexpensive system, the Chicago exchanges needed to replicate that as quickly as they could. So the CBOT lowered its fees practically to zero, to match the low fees being charged by Eurex U.S. And both the CBOT and the CME did everything they could to improve their electronic systems and to encourage existing traders to move their business out of the pits and onto the screens so that by the time that Eurex opened its new exchange it would be competing with largely electronic exchanges and not with floor exchanges.

So, while Eurex was dealing with the regulatory hurdles created by the Chicago exchanges, the exchanges took advantage of the breathing room to significantly upgrade their systems. The CBOT had been using a/c/e, a version of Eurex's own system, but was able to migrate its products onto the superior LIFFE Connect platform, so by the time Eurex U.S. was able to open its doors it was at a technological disadvantage. Those who worked on the CBOT and CME upgrades were impressed with the military precision with which both exchanges stuck to their timelines. They had to be on time because the Eurex U.S. launch was imminent. The CBOT completed its migration on January 2, 2004. Eurex U.S. turned on its screens on February 8, 2004, a little over a month later.

It was impossible to become a totally electronic exchange overnight, but a substantial amount of CBOT business was moved onto the screen before Eurex U.S. opened its doors for trading in early 2004. In fact, to the amazement of many observers, the strategy worked. Eurex failed to achieve any substantial share of the U.S. Treasury futures market. The CBOT was able to hang onto its monopoly. When it was clear to Eurex that the attack against the CBOT was going to be unsuccessful, it switched its attention to the CME currencies. Again, the CME had been able to shift such a large portion of its currency trading to the screen that Eurex did not have the advantage that it once had when it stole the German government bond from LIFFE in London.

Though Eurex failed in its attempt to make inroads into the U.S. derivatives space, the competitive threat did wonderful things for the industry. It caused the exchanges to lower their fees. It caused the exchanges to substantially increase the percentage of their trading that was done on screen. So, after almost two decades of playing around with technology on the periphery, the Chicago exchanges moved toward screens in a big way, pushed mainly by one of the biggest competitive threats that they had ever faced in their entire histories. Eurex had done a big favor for Chicago, even though it failed miserably at establishing its own beachhead and had to retreat to Frankfurt with nothing but a lot of bills for the abortive attack.

Chicago has been an electronic backwater compared to the rest of the world, but one good thing that can be said about the city is that it was way ahead of New York. There have been many derivatives exchanges in New York, but in the past decade they were all merged into two: the New York Mercantile Exchange (NYMEX) and the New York Board of Trade (NYBOT).

NYBOT was a bit of a contradiction. When the World Trade Center was attacked on September 11, 2001, NYBOT was the only exchange that had a physical backup location, and it reopened its trading floor at this Long Island location three days after the attack. However, for an exchange with internationally oriented contracts, NYBOT was surprisingly backward when it came to technology. In fact, it was the only one of the top four U.S. exchanges that had not even set up an after-hours electronic system. It had planned well for a physical disaster but had planned poorly for a competitive disaster. The lesson of an electronic exchange stealing the best product from a floor-based exchange was known to all from the DTB-LIFFE competition we’ve already explored. NYBOT was very much at risk of a screen-based exchange such as ICE, Eurex, Euronext, or the increasingly electronic CME or CBOT picking off its best products.

ICE was the one that made the play for the NYBOT contracts, but it didn't just list coffee, sugar, and cocoa and use marketing and various incentives to win over customers. Instead, ICE bought the products, customers, and liquidity by purchasing NYBOT outright. In this case, it was more than just the products. ICE had been outsourcing clearing, and with the NYBOT purchase, overnight it had its own clearinghouse. It wasn't too bad a deal for the NYBOT members; on the day they voted 93% in favor of the deal, it was worth $1.57 billion ($400 million cash and 10.297 million ICE shares). Divided among the 970 memberships, that came out to $1.6 million per seat, roughly twice the value of four months prior. ¹⁴

But NYBOT's failure to create its own electronic platform did result in the virtual disappearance of NYBOT as a brand as its products and clearinghouse were simply absorbed into ICE Futures, ICE's futures subsidiary. It could have been both better and worse. It could have been better if NYBOT had created a viable electronic platform several years before and benefited from the additional trading and revenue that electronic access seems to bring about because of the rise of algorithmic trading. Had NYBOT shifted to screens, demutualized, and launched an initial public offering (IPO), the value created for members could have been much better than what they actually ended up with.

But NYBOT's fate certainly could have been worse. If ICE or another exchange had simply listed the NYBOT products and successfully won the business away from NYBOT, the exchange and its members would have been left with virtually nothing except for a clearinghouse they could sell. So NYBOT should probably consider itself lucky, or more accurately, the former NYBOT members should consider themselves lucky, since there is no longer any NYBOT to consider anything.

NYMEX had moved much further down the electronic path than NYBOT. It first created an after-hours system called ACCESS back in the 1990s. It then upgraded that to a better system called Clearport in 2002, and it had begun using the system during the day for a number of products. But NYMEX had two problems. First, the system was not sufficiently robust to be the workhorse for all of NYMEX's products. Second, NYMEX squandered a lot of time and resources on pursuing a floor-based path in 2005 and 2006. This quixotic venture involved setting up, as previously mentioned, trading floors in both Dublin and London. The original plan for its new exchange in Dubai had it as a floor-based exchange, and it also planned to launch trading on the abandoned SIMEX floor in Singapore.

Having an inadequate electronic system and an unfortunate diversion of management attention devoted to pushing a floor-based expansion into Europe, the Middle East, and Asia left NYMEX highly vulnerable to attack from an electronic exchange. The first such attack came from the most logical place—ICE Futures Europe, the former International Petroleum Exchange, which had arguably become a U.S. exchange, even though it was still regulated in the United Kingdom. ICE Europe, like all the ICE subsidiaries, was lean, mean, innovative, and able to under-price. In February 2006, ICE listed a cash-settled clone of NYMEX's cornerstone product, West Texas Intermediate (WTI) crude oil. To add insult to injury, the cash settlement price ICE chose to use for settling its contract was the most liquid one around: the NYMEX futures price. So ICE didn't really want to kill the NYMEX contract, or at least not do so too quickly. It needed that NYMEX price in the same way that a parasite needs its host. If the host dies, the parasite is out of business.

Things did not look good—and then they got much worse. As ICE's market share for WTI crude oil edged up toward 30%, ICE also launched clones of NYMEX's gasoline and heating oil contracts. Then the CME announced that it was developing its own energy contracts. The born-again largely electronic CBOT decided to launch electronic gold and silver contracts, the major products in NYMEX's metals division. NYMEX's very survival was at stake. NYMEX saved itself only by jumping into the arms of one of its potential competitors—the CME. NYMEX's only choice was to become electronic immediately. LIFFE had lost the bund eight years earlier, in part, because it took too long to move to the screen, and by the time it did, DTB had 100% market share. The only way to become immediately electronic (that is, within a few months) was to outsource, and the only viable quick outsource vendor was the CME's GLOBEX. While the price of renting space on GLOBEX was not made public, NYMEX was not in much of a position to bargain, and it surely paid the price for not being ready with its own system.

Once it was up and running on GLOBEX, NYMEX was able to start taking market share back from ICE on energy and from the CBOT on metals. To help matters, when the CME acquired the CBOT in 2007, it was not able to bring CBOT's gold and silver onto GLOBEX along with the other CBOT products because of CME's commitment to NYMEX. So, NYMEX was pulled back from extinction by the CME, but as explained later in this book, the exchange ended up being acquired by that very same protector.

The first stock market to move in the direction of screen-based trading was not one of the richer, more sophisticated exchanges such as the New York Stock Exchange or the Tokyo Stock Exchange or the London Stock Exchange but rather the relatively disorganized OTC market in the United States.

The OTC market, of course, was the marketplace for stocks that did not meet the listing standards of the exchanges. It was a dealer market, where from one to 10 or more dealers would make a two-sided market in OTC stocks. Each dealer would quote a price at which he was willing to buy and a price at which he was willing to sell. If a customer told her broker she wanted to buy 1000 shares of a specific OTC stock, the broker would call several dealers to get their quotes and then chose the lowest offer of the three dealers.

Of course, shifting the OTC stock market to screens didn't simply happen but rather was instigated by Congress. It all started because the SEC became concerned about pricing abuses in the OTC market. Because there were a number of brokers trading OTC stocks, there were always different bid-ask spreads available in the market from different brokers for the same stock, and it was virtually impossible, or at least very difficult, for investors to know the best price for a given OTC stock. So the SEC decided to conduct a Special Study of the Securities Market, which it delivered to Congress in 1963. The study argued that the best way to improve price in the OTC market was through automation, and it charged the National Association of Securities Dealers (NASD) to solve the problem by building an automated quotation system whereby its market makers or dealers would post their bids and offers.

The new system, which went live in 1971, was called the National Association of Securities Dealers Automatic Quotation System, or Nasdaq. And it was what its name said—simply a quotation system. To actually buy or sell required a phone call. The only thing automated was the quotes. But this was a positive shift in transparency. Initially only the median bid and offer was displayed for 2500 OTC stocks. It actually took till 1980 before the best bid and offer was displayed on the screen. That was huge and resulted in a significant narrowing of spreads. Now all dealers and brokers could see the best dealer quotes for each OTC stock.

Actual electronic matching began in 1984 with the Computer Assisted Execution System (CAES) but leapt forward the following year with the 1984 introduction of the Small Order Execution System (SOES). However, this was still a dealer market, so SOES was not matching incoming customer orders with each other but rather with the dealer offering the best quote. Furthermore, SOES was not automatic—dealers had the option of responding or not. So, during the crash of October 1987, many small customer orders delivered via SOES were not accepted and went unexecuted. As a result, SOES participation became mandatory for dealers in 1988. Then came a long list of improvements at Nasdaq, including SuperSoes in 2001, which, among many other things, lifted the minimum size of a SOES order to 999,999 shares, and SuperMontage, a system that displayed and allowed execution against the best five bids and offers for every Nasdaq stock on a single screen. The ECNs played a very important role in stimulating this market improvement. Because they were competing with Nasdaq and with each other for order flow, the ECNs were continually improving their systems.

The situation was really Nasdaq's fault. The ECNs were mostly created after the SEC wrote new order-handling rules in 1997. And why did the SEC write new order-handling rules? Because a couple of academic economists, William Christie and Paul Schultz, published a clever article in the Journal of Finance¹⁶ (but released the results to the press prior to publication) essentially pointing out that Nasdaq market makers were colluding to earn bigger bid-ask spreads. There were none of the usual tools used to uncover cheating and collusion—no hidden wires, no bugs on phones, no subpoenaed emails—just statistics. Christie and Schultz showed that though the Nasdaq market traded in eighths of a dollar, an archaic thing to begin with in a largely decimalized world, most quotes and trades were done on the even eights, that is, 2/8, 4/8, 6/8, and 8/8 (i.e., a dollar even). If quotes and prices were random, you'd expect about the same percentage of quotes and trades to occur on the odd eights, but the dealers avoided the odd eights like a plague. The effect was to preserve wide 25-cent bid-ask spreads. To put this in the context of today, there are many stocks with bid-ask spreads of 1 cent, so the Nasdaq dealers were engineering spreads 25 times the size of those today. Both the U.S. Justice Department and the SEC got on the case, and the SEC wrote new order-handling rules, which essentially gave birth to the ECN craze and brought a new level of competition to the market. As part of the settlement, Nasdaq agreed to new order-handling rules and to integrate with other ECNs (also known as alternative trading systems). Thus the ECNs played a major role in bringing electronic trading to the equities market in the United States.

In the mid-1990s, the SEC gave ECNs the status of alternative trading systems and permission to integrate with Nasdaq's system. A number of ECNs, such as Instinet, Archipelago, Brut, and Island, entered the equities marketplace. ECNs provided market data (Level 1 and Level 2¹⁷) to the marketplace beyond traders and market makers at NYSE and Nasdaq. Boosted by the Internet and technological advances in the 1990s, the ECNs evolved into real competition for the two exchange giants. They brought transparency, efficiency, and anonymity to the equities markets and forced the NYSE and Nasdaq to reinvent themselves. By forming partnerships with various discount brokers and investments banks, ECNs made it easier and cheaper for the general public to trade equities. ECNs partnered with online brokerage houses such as E*Trade and Charles Schwab to generate liquidity. By partnering with other ECNs, they linked their order books to provide consolidated liquidity for the whole market. They challenged the rules¹⁸ of the primary exchanges and forced the NYSE and Nasdaq to rethink their traditional trading model for the first time. These markets, cognizant of the threat of ECNs, evaluated their option of demutualization, ¹⁹ a trend that was already common in European equities market and even in U.S. futures markets. Most ECNs today have been acquired by one of the primary exchanges or by large investment banks, but their innovation and persistence forced the NYSE and Nasdaq to grudgingly plan their survival for the next century.

The competition for electronic trading really heated up with the emergence of ECNs. The competition from ECNs such as Instinet, Island, and Archipelago forced NYSE and Nasdaq, the two dominant U.S. equity exchanges, to rethink their business models. The ECNs challenged the roles of Nasdaq dealers and NYSE specialists, who served as the middlemen for trade matching. Instead, ECNs matched trades automatically on their systems. The ECNs brought anonymity, lower execution costs, and detailed market data to the trading community. Anonymity, for example, has attracted more institutional traders to ECNs. On the floor of the NYSE and in the Nasdaq dealer community there was always a risk of word leaking out about a large order coming in, which could cause the NYSE floor traders or Nasdaq dealers to drive up the price. These leakages do not happen on a computer screen of ECNs where buyer and seller orders are matched by the system. ECNs' lower execution fees, for example, saved institutional investors 1–1.5% on their trades. So, on a $100,000 trade they would save $1000 to $1500. ²⁰ The two market giants looked to purchase the ECNs to maintain their grip on market liquidity. Most ECNs were eventually purchased by these exchanges²¹ and by some of the larger investment banks. With the acquisition of the ECNs, the exchanges were able to adopt just enough technology to improve their floor-trading but not eliminate the trading flow.

The major transformation of the U.S. equities markets did not occur until recently and only when faced with global competitors and the success of markets in Europe. Nasdaq used the infrastructure acquired from ECNs to provide an electronic trading platform to its customer base. ²² However, the NYSE continued to resist the full transformation to electronic trading. It acquired Archipelago on March 7, 2006, and became a public company. Similarly to CBOE's response to the ISE threat, the NYSE rolled out its hybrid system in October 2006 to quell the pressure of electronic trading. The hybrid system would allow electronic trading to exist in parallel to floor-based trading at the NYSE—and it fared well for a while. According to the exchange, the number of trades executing electronically tripled after the launch of the hybrid system. ²³

However, a year later, the exchange once again was facing challenges. The hybrid system in October 2007 was only matching 40% of NYSE listed volume, significantly lower than in the past. ²⁴ New waves of competition from new upstarts such as BATS and Direct Edge as well as from its old rival Nasdaq have continued to challenge the NYSE with their high-speed trading systems and has forced the NYSE to rethink its strategy once again. Today, Nasdaq has expanded globally, with the recent mergers with OMX, the Nordic exchange, and a deal to cooperate with Borse Dubai. The NYSE also merged with Euronext, creating the first transatlantic exchange, and continues to face pressures to complete its migration from floor to screen.

But as we can see from Table 3.2, though Nasdaq was an early adopter of technology, it took 30 years before it gave up the telephones altogether and replaced them with an electronic matching engine for all shares. In fact, the pattern with stock exchanges is similar to that with derivatives exchanges; the exchanges that converted early to full electronic trading were not in the world's financial capitals but rather in more out-of-the-way places like Busan, Santiago, Helsinki, Shanghai, and Riyadh. Furthermore, as in the case of the derivatives exchanges, U.S. exchanges were not leaders but rather laggards when it came to a full adoption of electronic trading.

There were other early pioneers, even though they didn't go fully electronic for some time. An important early player in the shift from floors to screens was the Toronto Stock Exchange (TSE). In 1977, it developed a system called CATS, a nice acronym for the more prosaic Computer-Assisted Trading System. CATS was first used for less liquid stocks. It was not uncommon to use electronic matching for marginal situations, such as less liquid stocks in Toronto and Tokyo, or for after-hours trading in derivatives at the CME, CBOT, NYMEX, and LIFFE, or for small orders, as in the DOT and Super DOT system at the NYSE. The TSE has since become a part of a larger exchange group called the TSE Group and uses a new platform, TSX Quantum; in 2008 it announced a new separate order book, TSX Infinity, for high-velocity traders.

The Australian Stock Exchange, which was created as an association of six different stock exchanges in Australia, was also an early adopter. In 1985 a committee from the exchange took a two-week tour of North American exchanges and came away most impressed with Toronto's CATS system. ²⁵ They came home wanting to use CATS as a model but to build something much faster and more advanced. They spent $1.5 million over the next few years developing the system they called SEATS, for Stock Exchange Automated Trading System.

From a regulatory point of view, it was clear that screen-based trading was highly superior. It created a very transparent audit trail. For example, one of the scams that existed on the Australian floors was something known as the top pocket account. The broker would execute a trade and then put the executed order in his pocket for later allocation. If the market moved up in a significant way, he would execute another trade on behalf of the customer and allocate the first trade to his own account, thereby earning a tidy profit. SEATS would prevent that from happening.

But the exchange got huge pushback from the floor traders, who feared they would lose their edge by no longer being able to make eye contact and read body language. The exchange engaged the members, gave a demonstration of the new system, took some 400 suggestions for changes, made the appropriate changes, held roughly 100 meetings to demonstrate the new system, and then was ready for final rollout. The exchange chose October 19, 1987, as the launch date for the new system for a portion of the listed shares. Yes, that October 19. Of course, on that day all the press the exchange invited to view the wonderful new system became suddenly uninterested in the clever technology. They wanted to find out why the market was dropping like a rock. It was much worse on Tuesday, when the major index, the All Ordinaries, fell 25%.

But the system passed the test, and the exchange stayed open even though neighbor Hong Kong closed both its stock and futures exchanges as a result of the crash. SEATS traded a limited list of stocks side by side with the floor for the next three years. In September 1990 all stocks began to trade side by side. By the end of the month, all six exchanges that make up the Australian Stock Exchange had closed their floors. The system did so well that soon after, the Swiss purchased Australia's system.

In some cases the transition was not directly from floor to screen but rather from floor to telephone to screen. When the London markets experienced—or suffered, as might be said by the thousands of workers who lost jobs—the Big Bang of 1986, the London Stock Exchange went electronic in a Nasdaq sort of way. In fact, its electronic system was called SEAQ, for Stock Exchange Automated Quotations system, and was, as its name suggests, a network allowing competitive market makers to post bids and offers on a screen. Because there was no electronic matching system, the actual trades were done via telephone. In principle, the trades could also be done on the floor, but the floor was quickly abandoned and all trading moved to the telephone, just as on Nasdaq at that time. So, in a relatively short period of time, one of the major exchanges of the world essentially became an OTC market. ²⁶

Despite the fact that the new system did not involve electronic trade matching, the combination of the new system along with other market changes resulted in a large increase in trading volume. The other changes included the removal of barriers to entry to the brokerage business from banks and foreigners and the elimination of fixed commissions. Institutional commissions quickly fell 50%, but turnover increased sufficiently to overcome that drop and give the LSE firms an actual increase in income. ²⁷

Just over a decade later, in 1997, the LSE brought in a new electronic platform called SETS (Stock Exchange Electronic Trading Service). And, having developed the habit of making a big change every 10 years, in 2007 it was time for the newest LSE system, called TradElect. In fact, today TradElect manages everything, and within TradeElect is a series of different trading services that utilize the earlier two systems, SEAQ and SETS, as well as some new ones, including an International Order Book for depository receipts. The 20-year-old system, SEAQ, is used for fixed income and stocks on smaller companies and is still a quotation system with no electronic matching.