Date: April 16, 2004
Publication: Institutional Investor By: Richard Rosenblatt, CEO, and Joseph Gawronski, COO, of Rosenblatt Securitie

Thinking Outside the Black Box: Accurate Price Discovery And Certainty Of Execution May Just Be Compatible

The following special commentary was written by Richard Rosenblatt, CEO, and Joseph Gawronski, COO, of Rosenblatt Securities, an agency-only brokerage firm that is a member of the NYSE, ArcaEx and the major ECNs. Rosenblatt and Gawronski provide InstitutionalInvestor.com readers with an in-depth look the hotly debated issues of best price and speed of execution in the context of the listed market and current pressures to rework its structure. "Accuracy of price discovery and certainty of execution need not be mutually exclusive if the time is taken to re-examine the current paradigm," they argue. They make a case for considering a new paradigm: auto-guarantee.

The superiority of the New York Stock Exchange's price discovery mechanism relative to that of other markets is supported by both objective studies and anecdotal evidence.[1] Since the primary function of any market is to discover price (the momentary equilibrium of supply and demand at the time of a transaction), this advantage would seemingly position the NYSE well to combat its ECN competitors who are attempting to divert listed order flow. However, perception issues surrounding corporate governance at the NYSE and the specialist investigations, as well as customer demands for speed, have left the NYSE vulnerable.

In addition, the Securities and Exchange Commission's pronouncements on trade-through in its proposed Regulation NMS[2] make it clear that advocates of certainty and speed of execution over accurate pricing have been gaining a powerful base of support and the primacy of "best price" may be pushed aside. In fact, even the NYSE is seeking to greatly expand auto-execution opportunities with its recent proposal to drop the current restrictions on Direct +. This suggests that the NYSE itself has either bought into the wisdom of these changes or, at minimum, is willing to sacrifice best price in certain instances to ensure that it will be characterized as a "fast market" to stem potential inroads by competitors, which could result from anticipated changes to the trade-through rule.

We believe the SEC and many market observers are underestimating the importance of accurate pricing to the efficiency of our capital markets by choosing automatic execution and speed at the expense of price discovery. Ironically, a choice does not have to be made. There is no reason to assume that certain, speedy execution and accurate price discovery cannot live compatibly together.[3]

In fact, the efficiency and possibly the dominance of U.S. capital markets may hinge on our ability to come up with just such a combination.

A New Paradigm: Auto-Guarantee

Let us begin by changing the paradigm. Instead of auto-execution, consider the possibility of auto-guarantee, which by separating price discovery from trade discovery could move us down a pretty interesting road. How might an auto-guarantee work in practice? Here's a simple example. Assume a market where 5,000 shares is offered at $30. If an auto-guarantee order to buy 1,000 shares is sent to the NYSE, the buyer would instantly receive an execution with the guaranteed price of $30. An auto-quote function would then display a bid for the same amount of shares but at a price lower by some pre-determined minimum increment, for instance, $0.05, resulting in a new displayed bid of $29.95 for 1,000 shares. If, within say a five second period, the specialist, a floor broker, a DOT participant or a competing marketplace hit that bid and agreed to sell the stock at $29.95, an adjusted price of $29.95 would be sent to the buyer. If not, the buyer would be executed at $30 (no worse off than he would have been with auto-ex), or if an offer between $29.95 and $30 had been displayed within the five seconds, the buyer would again receive a better, more accurate price. If another buyer paid $30 for the 5,000 shares or the offer canceled within the five second period, a guarantor would have to supply the stock from its firm account.

Who would be willing to provide a guarantee for such trades? We believe specialists would. One of the unique characteristics of the NYSE is the specialist system, which provides an ever-present pool of capital to smooth out price movements.[4] Specialists make the bulk of their profits from principal trading rather than commissions. The incentive to put their capital at risk by providing a guarantee is simply that increasing the number of trading opportunities increases potential profits. The specialist would therefore be the ideal participant to provide the guarantee if an auto-guarantee product were implemented.

The marriage of instant execution and accurate price discovery offers the best of both worlds.a guaranteed execution without compromising price discovery and the opportunity for price improvement. The only real substantive philosophical objection to an auto-guarantee concept would likely come from those who advocate rewarding market participants who enter limit orders.[5]However, there are two strong counterarguments to this objection.one a market structure argument as to why limit orders should not be rewarded beyond some basic protections (e.g. the equal treatment of limit orders across market centers)[6] and the other, a practical reason why they need not be rewarded.

Rewarding Limit Orders Should Not Limit Price Discovery

With respect to the market structure argument, while limit orders are certainly valuable in that they can help draw liquidity to a marketplace and form a starting point in the price discovery process, they should not be rewarded at the expense of accurate price discovery, an absolutely essential component of market integrity. Let's return to our offer at $30 that attracted 1,000 shares. Shouldn't that order be rewarded for exposing itself and attracting another order? The answer is yes.and no. The reward to our $30 offer should be an opportunity to trade, but not the right to trade at a poorly discovered price. Even if it were the possibility of trading at an advantageous, inaccurate price that incented our seller to place the limit order, there's more to consider. Remember accurate price discovery? For markets to be efficient, they must provide a system which encourages accurate pricing.

The big drawback to auto-execution is that by its nature it prevents the trading opportunities necessary for accurate price discovery. The offer at $30 is the starting point of price discovery, not necessarily the end point. Auto-guarantee, or some similar product, instead creates a new pricing possibility by displaying an improved bid. The most efficient pricing mechanism is one that maximizes trading opportunities while maximizing the sunshine on the process, making the opportunities available to all. The "black box" is properly named. Not only does auto-execution inhibit any trading opportunities other than the initial one created by the limit order, but it does so in total darkness.

Market centers that incent limit orders with auto-execution have badly missed the point. The incentive they offer is the premium received by the limit order in the form of mispricing, the difference between the auto-execution price and the price that would have been reached if the price were permitted to have been discovered competitively. While it is understandable for investors and their brokers to seek advantageous mispricing, it is absurd to think that the market or the SEC should be an accomplice. The NYSE, like all markets, must continue to seek the most accurate pricing that best represents supply and demand. The market with the most trusted price will attract, as it should, the most liquidity.

The "Red Herring" Of Rewards

Practically speaking, the professed need to reward limit orders in order to increase liquidity is a red herring. To begin with, limit orders historically have never been used to represent actual available liquidity. Instead, with the exception of small retail orders, limit orders displayed in any market have always been tactical advertisements designed to attract liquidity. Limit orders do help create trading opportunities in their willingness to be exposed, but this is simply a tactic. When the tactic is successful, the limit order gets a trade (and often a much larger one than the displayed limit order once the contra party has been drawn into a negotiation). When the tactic is not successful and the person posting the limit order has not properly priced the stock, he does not get a trade, nor should he. Supply and demand did not meet at that price point. The market with the deepest pool of liquidity will receive the most limit orders without the need to offer any additional reward to these limit orders beyond the greater likelihood of trading that the inherent liquidity on that market provides. The benefit to a limit order, or any order, is the resultant increased opportunity to trade. To add a pricing incentive through automation, at the expense of the other side of the trade and ignoring the market's responsibility to price accurately, is completely unnecessary.

The example of ECNs is instructive in this regard. ECNs reward limit orders by providing both an unfettered auto-ex capability and an explicit financial incentive in the form of a rebate for posting limit orders (currently $0.20 per hundred shares on most ECNs). Despite these incentives, the size of the limit orders displayed on ECNs is typically small, and traders utilize reserve features or active management of their orders rather than larger limit orders to trade larger size. The reason is rather simple. Exposing size poses a risk in the form of potential market impact by attracting competitors rather than counterparties. Are institutional investors really going to post limit orders representing tens or hundreds of thousands of shares and risk moving the market by exposing their intentions just because they can no longer be "pennied"? We believe it is naï to think so. Rewarding limit orders by hampering the ability of other market participants to offer price improvement will not result in larger limit orders being posted or increased liquidity, but rather will simply result in weaker price discovery.

Accuracy of price discovery and certainty of execution need not be mutually exclusive if the time is taken to re-examine the current paradigm. A product like auto-guarantee, and a clear understanding that a market must support best price for all order types, offers a unique opportunity. The historic leader in accurate price discovery through price improvement is the NYSE. With an auto-guarantee product, the NYSE could well emerge as the leader in automated markets as well.

[1] With respect to objective studies, see the SEC's Report on the Comparison of Order Executions Across Equity Market Structures, January 8, 2001 (concluding that average effective spreads were significantly wider in Nasdaq trading than the NYSE in all comparable stock categories other than $50 billion plus market caps where they were nearly the same); and Birinyi Associates' Bulletin: Trading Costs dated March 4, 2004 (concluding that in December 2003 trading cost investors 5.06 basis points in impact on the NYSE and 6.53 basis points, or 8.81 basis points if certain apples-to-apples adjustments are made, on the Nasdaq 100, using a methodology outlined by the SEC in 1971). With respect to anecdotal evidence, on recent visits to three clients that run quantitative portfolios each of the clients confirmed that execution quality on the NYSE was better than what they were experiencing on the Nasdaq.

[2] SEC Release No. 34-49325; File No. S7-10-04. For a detailed examination of the inadvisability of trade-through rule repeal and/or weakening, please also see Debunking the Myths of the Trade-Through Rule: A View from an Agent.

[3] Incidentally, there is also no reason to assume, as most do, that ECNs provide speed and certainty of execution. Using publicly available SEC Rule 11Ac1-5 data, Celent Communications recently concluded in its March 11, 2004 report entitled Execution Quality in Equity Trading: Electronic Communication Networks that "Speedy executions are not as speedy as expected from electronic trading platforms. Marketable orders on some ECNs wait as long as 3-5 minutes for execution....Certainty of execution on ECNs is not so certain. Averaging the execution rates of all market participants included in this report series (including ECNs, specialists, market makers, institutional and retail brokerages, among others), 83% of marketable share volume is executed. In contrast, ECNs execute only 39% of their marketable share volume."

[4] A study released in December 2003 by the NYSE's Chief Economist supports the view that the NYSE suffers from less volatility than the Nasdaq with empirical evidence. The study examined the stocks of 39 companies (average market capitalization.$1.5 billion) that voluntarily transferred from the Nasdaq to the NYSE during the period January 2002 to March 2003. The study documented a significant drop in volatility for these stocks.

[5] See, for instance, Testimony of Gus Sauter, chief investment officer of The Vanguard Group, before the Subcommittee on Capital Markets, Insurance and Government Sponsored Enterprises, Committee on Financial Services, United States House of Representatives, Market Structure III:The Role of Specialist in the Evolving Marketplace, February 20, 2004. While we disagree with his conclusions about the reward that limit orders should receive, his testimony is nevertheless a thoughtful explanation of the philosophical basis for protecting limit orders and we certainly agree with his statement that "There is no need to debate whether best price or speed and certainty is better. Investors require both...."

[6] In Regulation NMS, the SEC has properly focused attention on the need to give some basic protections to limit orders in the form of a proposal for a uniformly applicable trade-through rule, but also makes clear that it does not advocate protecting limit orders in an absolute sense since it has proposed opt-out capability and de minimis exceptions to the trade-through rule.

< Back to Press page >