The comments contained in this article reflect solely the opinions of UBS Warburg's EMM unit. They have NOT yet been approved by the bank as a whole, and should not be construed as UBS Warburg's position on these issues.
I enthusiastically applaud the Commission's effort to encourage greater degrees of transparency and disclosure in the US equity market. I agree with the Commission that transparency and disclosure are the foundation of fair competition. In order to achieve this goal, one must be concerned with the fact that the market players the proposed Rule 11Ac1-5 and 11Ac1-6 are meant to regulate will be highly motivated to find creative ways of evading the spirit of the rules while following the letters. I think the following areas are particularly worthy of the Commission's attention:
1. The benchmark for any price improvement statistics is almost certainly the NBBO (National Best Bid and Offer). Unfortunately, NBBO is by no means an accurate measure of where liquidity resides because several market centers have serious speed limits on incoming orders. For example, if a 5-second delay exists between order initiation and the destination market center's acknowledged reception, the quote is effectively always 5 seconds out of date. Therefore, this delay is a very important measure of execution quality and should be incorporate into the disclosure requirement. Furthermore, the failures to honor one's own quotes should be penalized more seriously than the failures to match the NBBO posted elsewhere.
2. Currently, the US equity market place is quote-driven, as opposed to the order-driven markets in Europe and Asian. This fact leads to serious ambiguity in what constitutes the best execution or price improvement in a system with multiple market centers. For example, if the national best bid is 200 shares @ $50 posted at market center A, followed by 5000 shares @ $49.50 at market center B. Customer C sends dealer D a market order to sell 5000 shares. What is the minimum price that qualifies as the best execution if D wants to take on the trade as principal? If D buys at $50, he clearly has to supply far more liquidity than the rest of the market is willing to. If D buys at $49.50, however, C can argue that at least 200 shares of the trade should receive a better price. If D sends 200 shares to A, he risks facing a long delay as well as the possibility of rejection, not to mention the danger of adverse price movement during the wait and a bad execution speed statistics. A fair solution to this dilemma, which is basically caused by the uncertainty in the next-to-best orders in limit order books, is to require full transparency on all limit order books. In the previous example, if market center A has another limit buy order of 300 shares @ $49.70, dealer D, who wants to offer price improvement, can simply break up the original order and send a market sell order of 200 + 300 = 500 shares to A, and fill the rest himself @, say, $49.60. Of course, the benefit of having transparent limit order books does not stop here. It also prevents large limit orders from being obscured by small ones. This should be an even more important consideration in the decimalized world.
3. It is well known among social scientists that "numbers can be tortured into saying anything." Summary statistics, no matter how carefully specified, often leaves room for unorthodox definition and/or interpretation. Surely the Commission does not want to be forced into the role of a forensic auditor over the gigabytes of raw data generated each day just to police the accuracy of the summary statistics derived from these data. It would be much better to simply require the disclosure of raw order execution history (excluding the identity of traders) and leave the policing task to independent auditors and academic researchers. This proposal truly adheres to the spirit of public disclose. It is also a market-oriented solution that minimizes the regulators' burden and frees up their resource to focus on clear-cut compliance issues such as whether the time stamps have been accurate.
4. Another drawback with pre-designated summary statistics is that they often do not tell the whole story. As a result, certain market centers may be favored over others. For example, exchanges with large market shares and/or slow execution speed have a much greater chance of matching customer orders in between quotes. Therefore they will automatically score higher in price improvement statistics without any effort on the part of their market makers. While some may argue that dominant and slow market center offer natural advantages, I strongly disagree. To be more specific, let's consider a hypothetical dealer whose business model is to randomly select a portion of incoming market orders and offer real price improvement, while the rest of the market orders are redirected towards the dominant market center without any delay. Clearly, the presence of this dealer improves the liquidity supply and execution quality for his customers with no downside whatsoever. His price improvement statistics, however, may look worse than the dominant market center simply because the destination can impose delay on incoming messages, and an erroneous conclusion may be drawn that it is better for the customers to send their orders to the dominant market center directly. For this reason, I again urge the Commission to adopt the alternative version of the proposed rules, where the entire order execution history is made public.
5. Lastly, I would like to point out that slow execution speed often gives a market center the opportunity to reduce its risk at the expense of the customers. To illustrate this point, let's consider the following example. Suppose both market centers A and B are quoting 1000/1000 shares at $50.00/$51.00. Both receive market orders to buy 1000 shares. B fills the order immediately at $50.50, but A takes his time. If the prices move to $48/$49, A fills the customer at $49. If the prices move to $52/$53, he fills at $52. Both market centers would have provided identical price improvement of $0.50 on average, but A has no downside risk whatsoever while B is fully exposed to the price fluctuation. Since one of the main purposes a market maker serves is to provide immediacy and take on customers' risk, market center A is shirking his responsibility while still reaping the rewards. Hence, if the Commission chooses to require only summary statistics, the relevant benchmark for price improvement should therefore be the best NBBO during the period between order reception and execution, instead of the NBBO at reception. In the earlier example, if the prices move to $48/$49, the benchmark would be $48 instead of $50. Market center A would have an average price disimprovement of $0.50 compared to B's improvement.