How Any Market Revamping Might Remove, Add Wrinkles

By GREG IP, Staff Reporter of THE WALL STREET JOURNAL, Feb. 29, 2000

The structure of the entire U.S. stock market could soon change, affecting all investors. And the potential shift is aimed partly at dealing with controversial scenarios like this:

A stock you like is offered on your screen for $20.25. You don't want to pay that, so you give your discount broker an order to buy 500 shares at $20.125. Your bid pops up on the screen: It's now the highest in the market, and you figure it's only a matter of time before you get your stock.

Then you see 500 shares trade at $20.125. Then another block at that price. And another. Still you don't get any stock. Then your own broker buys 200 shares from another investor by paying $20.1875, a smidgen above your bid.

The stock rallies, and your broker unloads the shares he just bought. Frustrated, you cancel your first order and enter another with no preset limit. Five seconds later, your broker confirms you bought 500 shares at $20.375. You paid $125 more than you expected, but at least you paid just $10 in commission.

These things routinely happen to investors big and small, in both Nasdaq stocks and New York Stock Exchange stocks. It goes on because U.S. stock trading is distributed among the stock exchanges, dealers and electronic trading systems known as ECNs.

"Fragmentation" is now being hotly debated by Wall Street, the Securities and Exchange Commission and Congress. Tuesday, the heads of some of Wall Street's biggest firms plus the NYSE and the National Association of Securities Dealers, Nasdaq's parent, will testify on the issue to the Senate Banking Committee, whose chairman is Phil Gramm (R., Texas).

It's an issue that arouses considerable passions. Critics of fragmentation refer to it as balkanization. They say the solution is a so-called central order book into which orders from every exchange, dealer and ECN are funneled, where the first order to arrive is the first to be executed.

But many dealers and ECNs believe fragmentation is just another word for competition. Without that competition, they argue, you wouldn't have $10 commissions, five-second executions and many other investor benefits, such as trading after hours.

Whether fragmentation is a problem is "somewhat in the eye of the beholder," says Robert Schwartz, a finance professor at Baruch College in New York. "There are benefits, competition-wise, to fragmentation, and there are benefits in terms of quality of markets to concentration."

The empirical evidence suggests central markets do consistently provide better prices, says Dan Weaver, another finance professor at Baruch, but if some protections are put in place, fragmentation can help keep bid-offer spreads narrow.

More broadly, fragmentation presents what economists call the "free rider" problem. Investors willing to post the best bids or offers aren't necessarily those whose orders get executed. Instead, other dealers simply use that investor's bid or offer as a reference price at which they will execute their own customers' orders.

The SEC, in a release on the subject last week, said this means "vigorous quote competition may go unrewarded." If a customer improves the market but it's other investors who benefit, where's the incentive to improve the market?


How this happens varies on Nasdaq and the NYSE. On Nasdaq, dealers must post customer limit orders that improve the national best bid or offer. So if a stock is bid at $19.75 and offered at $20.25 and Dealer A gets a bid from Jim of $20, he must post Jim's bid, and the market will become $20 to $20.25. Alternatively, he might post Jim's bid on an ECN such as Instinet Corp. or Island ECN Inc. The result is the same.

Now here's the catch. Let's say Dealer B has a customer, Jane, willing to sell at the highest bid possible. Dealer B doesn't have to direct Jane's order to Jim's bid, because Jim isn't his customer. Dealer B doesn't even have to expose Jane's order publicly to see if someone else will pay more than $20. Rather, he buys it himself at $20. Then, hopefully, a buyer will come along to whom Dealer B can sell the stock at $20.25, for a 25-cent-a-share profit.

This practice of executing orders internally using the public best bid or offer as the benchmark is called "internalization" and is how all Nasdaq dealers have traditionally treated their own customers' orders.

Now, a huge share of Nasdaq's volume comes from customers of online brokers, who direct orders in particular stocks exclusively to wholesaler dealers such as Knight/Trimark Group Inc. and Charles Schwab Corp.'s capital-markets unit. Wholesalers usually pay for other brokers' order flow because of the opportunity to trade profitably. Payment for order flow is another form of internalization that is under debate.

The NYSE also has a free-rider problem in that regional specialists and nonmember dealers use its prices to execute orders they buy from other brokers. This is less of a problem than on Nasdaq because more than 80% of NYSE-listed stocks are executed by the NYSE specialist. But NYSE members are gaining new freedoms to internalize.

The SEC has suggested some ways to ensure investors who post aggressive prices are rewarded for it. But customers may still suffer. Suppose a Nasdaq dealer or a Big Board specialist has many customer-sell orders at $20.25. He might think that's bearish for the stock. So when a buy order arrives, the dealer "steps ahead" of those customers by offering at $20.1875. The buyer got an improved price but the sellers got nothing. The expected switch to pennies will make it even cheaper for specialists and dealers to step in front of customers.

A central order book would eliminate these problems by funneling all orders to one place, and by giving everyone, not just dealers or specialists, access to limit-order data. But some object that if dealers, ECNs or regional stock exchanges have to give their orders to a central book, they would have less incentive to compete for orders on price or service.

The New York Stock Exchange, for example, has sometimes been slow to innovate because its centrality overcame disadvantages of speed and cost. Only last fall, under a growing competitive threat, did it eliminate specialist commissions on small orders and propose its own ECN to speed up executions.


  1. Which problems will a central order book solve?
  2. Which problems will it not solve?
  3. Is there any basis for defending fragmentation?