The Nasdaq: How a 1971 Screen Network Remade the Over-the-Counter Market

On 8 February 1971 a dealer in Trumbull, Connecticut, switched on a network of terminals that displayed, for the first time, what more than 500 securities firms were willing to pay for some 2,500 over-the-counter stocks. A broker in Denver could now read a price that had been set in New York minutes earlier. What he could not do was trade on it. He still had to pick up a telephone.

That gap — between a price a screen displayed and a price a human being would honour — defined the next twenty-seven years of the market the National Association of Securities Dealers had just automated. It closed only after two academics counted fractions in a database, a federal antitrust division wrote a complaint, and thirty-seven brokerage firms paid out more than a billion dollars.

A Market Without Published Prices

Before the terminals, the over-the-counter market was quoted on paper. The National Quotation Bureau printed the pink sheets each morning: several hundred pages of dealer quotations for thousands of unlisted securities, distributed by mail, priced for the trade rather than the public, and stale by the time a customer saw them. A retail investor who asked what a stock was worth received a number from his broker and no way to check it.

Congress ordered an examination of this arrangement in September 1961, appropriating $750,000 for a study of the securities markets. Milton Cohen, a former Commission attorney, was hired to run it, and he insisted on a degree of independence unusual for a government inquiry — his group would report its findings to the Commission without requiring the five commissioners to approve them first. The resulting Special Study ran to roughly 3,000 pages and reached Congress in 1963 (Cohen, 1963). Its treatment of the over-the-counter market found quotations that were wholesale-only, fragmented across dealers, and impossible for a customer to verify.

Its recommendations produced the Securities Acts Amendments of 1964, which extended registration and periodic disclosure to larger unlisted companies. Quotation itself was left to the industry, and the NASD contracted the Bunker Ramo Corporation to build a machine that would collect dealer quotes electronically and push them back out.

Three Levels of Sight

What launched in February 1971 was a quotation display, not an exchange. Bunker Ramo's system sorted its users into tiers, and the tiering mattered more than the technology.

A customer's broker, at Level I, saw a summary. The dealer on the other side of the trade, at Level III, saw the whole book and knew exactly where his competitors stood. Screens had replaced the pink sheets without dissolving the asymmetry that made dealing profitable, and the NASD had automated the display of prices while leaving the negotiation of them on the telephone.

Growth came anyway. Real-time reporting of completed trades arrived for the largest issues in 1982 under the Nasdaq National Market designation, which meant that for the first time a customer could see what a stock had actually traded at rather than what a dealer said he would pay. Automated execution for small orders followed in December 1984, when the Small Order Execution System opened for 25 stocks and filled orders of 1,000 shares or fewer against a market maker's displayed quote.

Participation in SOES was voluntary. That detail went untested for three years.

The Phones That Went Unanswered

On 19 October 1987 the equity market fell by a fifth in a session, and the Nasdaq dealer network stopped functioning in a way the exchange floors did not. Market makers who did not wish to buy simply declined to pick up the telephone. Orders that a customer had placed in the morning went unexecuted through the afternoon, because the obligation to honour a displayed quote existed on paper and the mechanism for enforcing it did not.

Comparison with the specialist system that absorbed the same selling pressure in New York was unflattering, and the regulatory response was to remove the dealer's discretion. From June 1988 SOES participation became mandatory for market makers in National Market System securities: a displayed quote now had to be executed automatically for small orders, whether the dealer answered his phone or not. SelectNet, launched the same year, moved negotiation of larger orders onto the screen as well.

Dealers disliked the change, and a population of traders soon emerged to exploit it — firms that hit stale quotes on SOES faster than a market maker could refresh them, and who were known in the industry as SOES bandits. Their existence was an argument, though nobody framed it that way at the time, that the dealers' quotes were not as sharp as the dealers claimed.

The index the system had been given at launch, set at 100.00 on 5 February 1971, fell to 54.87 on 3 October 1974 — a closing low it has never revisited — and did not clear 200 until 13 November 1980. It passed 500 on 12 April 1991 and 1,000 on 17 July 1995. By the middle of that decade Nasdaq carried the listings of Microsoft, Intel, Apple, and Cisco, and it was routinely described as the market that had beaten the New York Stock Exchange to the future.

Two Professors Count Fractions

William Christie of Vanderbilt and Paul Schultz of Ohio State were not investigating anyone. They were examining how quotes behaved in a market quoted in eighths of a dollar, where a spread could in principle be one eighth — 12.5 cents — and were running the distribution of quoted fractions across a hundred of the most actively traded Nasdaq securities.

What they found was a hole in the data. Odd-eighth quotes — prices ending in one eighth, three eighths, five eighths, seven eighths — were, in their phrase, "virtually nonexistent" for 70 of those 100 stocks, among them Apple Computer and Lotus Development. Dealers quoted in even quarters. A spread that avoided odd eighths could not be narrower than a quarter of a dollar, twice the minimum the tick size permitted.

No agreement was needed to produce this pattern, and Christie and Schultz did not claim to have found one. Their paper stated that they could not reject the hypothesis that market makers in active Nasdaq stocks implicitly colluded to maintain spreads of at least 25 cents by avoiding odd-eighth quotes (Christie and Schultz, 1994). A convention observed by everyone and enforced by nobody produced the same arithmetic as a cartel.

Newspapers reported the finding on 26 and 27 May 1994. Within days, dealers in several of the affected stocks began quoting in odd eighths, and the spread on those securities halved almost overnight. Christie, working with Jeffrey Harris and Schultz, published a companion paper asking why market makers had stopped avoiding odd-eighth quotes; the answer, plainly, was that somebody had started counting.

The Convention Meets the Antitrust Division

Two investigations followed, one civil and one regulatory, and both accepted the professors' arithmetic.

The Antitrust Division filed a civil complaint under Section 1 of the Sherman Act on 17 July 1996 against firms making markets in Nasdaq securities. It alleged a "common understanding" among them to follow a "quoting convention" that had inflated the inside spread in certain stocks. Nobody was accused of a meeting or a written agreement — the complaint described an understanding that participants absorbed and enforced through the ordinary pressure of a dealer community whose members needed each other's quotes every day.

The Commission's own settlement with the NASD came three weeks later. Its report under Section 21(a) of the Securities Exchange Act, issued on 8 August 1996, found that Nasdaq market makers had adhered to a pricing convention that often increased the transaction costs paid by customers, and that the NASD had failed to enforce its own rules against the firms it was supposed to supervise. In place of a penalty, the NASD undertook to spend an additional $100 million over five years on regulation and surveillance. Arthur Levitt, announcing the settlement, treated the failure as one of self-regulation rather than of technology (Levitt, 1996).

The governance problem had already been diagnosed from inside. A select committee chaired by former Senator Warren Rudman, appointed by the NASD's own board in November 1994, delivered its report on 15 September 1995 and found that the association's structure had "blur[red] the distinction between regulating the broker-dealer profession and overseeing the Nasdaq stock market" (Rudman, 1995). One organisation ran a market and policed the dealers who made money in it. The committee recommended splitting the two functions, and the NASD did so, creating a separate regulatory subsidiary.

Private litigation ran longer and cost more. The class action against the market makers was settled for $1.027 billion, approved by a federal judge on 9 November 1998 and paid by 37 domestic brokerage firms — the largest civil antitrust recovery to that date.

Rewriting the Quote

Reform of the market's plumbing was the durable part. The Order Handling Rules, adopted in August 1996, took effect on 20 January 1997 for exchange-listed securities and a limited set of Nasdaq issues, and phased in across the rest of the market during that year.

Two provisions did the work. A customer's limit order priced better than the dealer's own quote had to be displayed rather than pocketed, which meant the public could now compete directly with a market maker instead of trading against him. And a dealer who posted a superior price in a private electronic trading system had to show that price in the Nasdaq quote as well, which ended the practice of running a tighter market for professionals than for customers.

Measured effects were large. The most cited study of the reform concluded that quoted and effective spreads "fell dramatically without adversely affecting market quality" (Barclay et al., 1999) — that the cost of trading fell without the loss of depth the dealers had warned would follow. Decimal pricing, completed across the American markets in 2001, cut the minimum increment again, and the electronic communication networks whose quotes the rules had forced into public view grew into the venues that now carry most American equity volume.

Nasdaq itself completed the journey from quotation network to exchange. The NASD spun it off, it registered as a national securities exchange, and the screen that had once only displayed a dealer's price came to execute against it. London had reorganised its own market around screens a decade earlier, abolishing its trading floor within weeks of switching its quotation system on, and the American automation that began with the back-office breakdown of the late 1960s arrived at the same destination by a slower route.

What the Screen Revealed

Nasdaq's founding premise was that publishing prices would discipline the people who made them. It half worked for twenty-three years. Quotes became visible, continuous, and national, while the convention that governed how they were set stayed invisible because no rule required anyone to look at the distribution of the fractions.

The market that Christie and Schultz examined in 1994 was the largest and most technologically advanced dealer market in the world, and it was quoting some of the most heavily traded shares in America at a spread twice the minimum its own tick size allowed. The companies that would define the next decade's boom were listed on it, and the fully automated markets that followed inherited its architecture of competing electronic quotes.

An index at 100.00 in February 1971 had become a symbol of American technological confidence by the time it crossed 1,000. The number nobody had thought to check was smaller: one eighth of a dollar, absent from a hundred stocks, worth a billion.