What Financial Trading Systems Tell Us About Markets

JULY 16, 2006

From Hans Stoll, Electronic Trading in Stock Markets, JEP, Winter 2006, 20:1.


1. Moving towards fully electronic market. NYSE has just merged with Archipelago and NYSE with Instinet.

2. Economies of scale and their affect on firm size and organization:

“Technology has changed the nature of the specialist in other ways, too. In 1975, 381 individual specialists owned seats and operated 67 specialist firms organized primarily in partnerships. Today, the number of individuals acting as specialists remains about the same, but they are organized into only seven specialist firms, structured as corporations. The consolidation of dealer firms on the NYSE (and also in other markets) reflects the economies of scale in the dealer business.” [p. 157]

Rmk: Might it also reflect the consequences of competition which has spurred the creation of a more oligopolistic market in order to preserve rents:

The bid-ask spread is a measure of the cost of immediacy, a term coined by Demsetz (1968). Dealers supply immediacy and earn the spread, while demanders of immediacy pay the spread. The spread compensates dealers for the risks they assume in buying (or selling) a stock that someone else is anxious to sell (or buy), as well as for the costs of processing the trade. The spread may also reflect monopoly rents if dealers have market power. [p. 156, emphasis added]

Illegal Activity by Specialists

Specialists are frequently criticized for the conflict inherent in their dual roles as brokers for the orders left with them to be entered in the book and as dealers trading for their own accounts. …

The Securities and Exchange Commission’s most recent investigation of specialists found that certain individual specialists traded ahead of customer orders in violation of the negative obligation not to trade “unless reasonably necessary”. The firms paid a total of $242 million to settle the charges against them (SEC, 2004). On April 12 2005, 15 individual specialists who worked for the firms were indicted, and the NYSE paid a fine of $20 million for failure to regulate the specialists properly. … [ed: goes on to list the details of these violations such as trading ahead of customer orders and interpositioning.]

The source of these problems is that the execution of trades is not automatic. The specialist has discretion … [pp. 158-159]

But …

Evidence of Collusive Behaviour in Nasdaq

Stoll states, pp.159-160:

In view of the number of competing market makers [2 or more compared to NYSE 1] and the apparent high-tech and transparent nature of the Nasdaq market, it might seem that transactions costs would be reduced with better prices for investors. However, all was not well. Several academic studies and investigations by the SEC and the Justice Department revealed dealer behavior that articifially raised bid-ask spreads above competitive levels (see in particular Christie and Schultz, 1994, 1995, Christie, Harris and Schultz, 1994, Huang and Stoll 1996). The source of the problem was that in a pure dealer market, which Nasdaq resembled at that time, customers must trade at dealer quotes. …. Consequently the bid-ask spread was determined by dealers without the possibility of competition from customers.

For several reasons, dealers did not compete effectively among themselves to narrow spreads. First, as Christie and Schultz (1994) emphasize, dealers seemed to coordinate their quotation patterns by only using price fractions that ended in even-eights … which meant spreads were at least $0.25. Second the practice of “preferencing” assured dealers they would receive order flow whether or not they quoted the best price. … Since much of the order flow in a stock was preferenced, a given dealer could not attract much additional order flow by improving quotes. Instead, business for all dealers (since each dealer committed to trade at the best quote). Consequently, the benefit to a dealer or improving the quote was low, and as a result, dealers limited their competition on quotes. Third while quote competition was limited on the public Nasdaq market, dealers did offer competitive quotes over interdealer trading systems in order to manage their inventory. These quotes were not available to the general public, however. The net effect of these factors was to raise the bid-ask spread above the competitive level, especially for smaller trades. Huang and Stoll (1996), for example, show that spreads in Nasdaq stocks in the early 1990s were twice those of comparable NYSE stocks. [pp. 159-160, emphasis added]


Given that the financial trading systems discussed above are, in many respects, the apotheosis of the market based exchange systems it is interesting to consider:

  1. The distance the markets are from textbook perfect competition (7 dealing firms on the NYSE, 1 market-maker per stock etc). It is also yet another example of the empirical bankrupty of the contestable market hypothesis.
  2. The potential for monopolistic or oligopolistic abuse was intensified by the ability to set institutional rules (especially on Nasdaq) along with market features conducive to such behaviour in the form of frequent and repeated interaction among participants (all of which facilitates abuse).
  3. Significant abuse did occur with substantial costs to almost all participants
  4. The crucial role played by external official (SEC, Justice Department) and unofficial (academics) regulators in correcting and sanctioning these abuses and promoting the overall efficiency of the market.1

To put it most starkly I would summarize the implications for policy as: efficient markets cannot exist without regulation

  1. “In 1993 the quoted half-spread on Microsoft exceeded 16 cents. By the end of 2001, the quoted half-spread had declined to under 2 cents. On Nasdaq, the most dramatic decline occurred as result of a May 1994 meeting of the Nasdaq administration with dealers to urge a reduction in spreads. The meeting was called in response to the Christie and Schultz (1994) paper and the emerging controversy over Nasdaq dealer behavior.” ↩︎