The proposal of the Securities and Exchange Commission for market structure reform, like so many other recent SEC initiatives, seems entirely ad hoc. There are certainly good reasons for considering reform of a market that consists of two separate and entirely different structures--one centralized and lacking any significant competition, and the other decentralized and highly competitive--but the SEC did not present any. Instead, the commission's proposal, which is not based on either market data or a vision of how the securities markets should function, seems designed to placate contending interests rather than meet the needs of investors or companies, now or in the future.
In late February, the Securities and Exchange Commission published Regulation NMS, a long-awaited proposal for market structure reform. There is little doubt that reform is overdue. The last major restructuring of the securities markets occurred in 1975, when Congress adopted a plan for a national market system. In that legislation, which followed serious SEC studies of the securities market in 1963 and 1971, Congress seemed to contemplate a number of regional securities markets that would compete with the New York Stock Exchange and a future in which investors would be able to access these markets without the use of intermediaries. This is not, of course, the structure that has since developed. Today, the NYSE remains the dominant market for NYSE-listed securities, many of the regional exchanges have withered, and investors who want to buy or sell NYSE-listed securities must still do so by placing orders with brokers.
The Ascent of Nasdaq
But while the structure for trading NYSE-listed securities has remained largely immutable, there has been a virtual revolution elsewhere in the securities markets since 1975. An informal dealer market that traded non-listed securities over the counter in 1975 became a formalized dealer market known as Nasdaq; entirely new computerized order-matching venues (known as electronic communications networks, or ECNs), arose to challenge the Nasdaq dealer market; and in response Nasdaq itself became a fully electronic market. In some ECNs, institutional traders can trade anonymously with other institutions without the intercession of a broker. In other words, the original vision of Congress for a securities market of competing venues and direct access by investors has been partially realized--but in the Nasdaq market, not in the market for NYSE-listed shares, where Congress apparently intended that it occur.
The reasons for this are complex, but SEC regulatory decisions are probably an important factor. Since 1975, the SEC has approved rules--such as the trade-through rule--that have tended to support the dominant position of the NYSE in trading NYSE-listed securities. The trade-through rule requires that orders to buy or sell securities listed on any registered stock exchange be sent for execution to the market where the best price is posted. Because the NYSE is the largest and most liquid market for these securities, it generally has the best prices, and thus the trade-through rule has reduced the likelihood that any serious competition for the NYSE would arise.
As originally framed, Nasdaq was a dealer market; it consisted entirely of dealers who made markets in securities that were not listed on any registered exchange by offering to buy or sell shares as principals. In this market, an investor (through a broker) would survey the prices and number of shares each market maker was willing to buy or sell and then choose the dealer or dealers with whom he or she would execute a trade. Since Nasdaq was a dealer market and not a registered exchange, it was not subject to the trade-through rule and was thus vulnerable to competition from a source that offered superior prices or services.
Such a competitor appeared in the late 1990s, when ECNs began to offer computerized matching of offers to buy or sell Nasdaq securities. Over time, as institutional investors found that they could effect trades at lower cost and achieve better overall pricing on the ECNs, these competitive trading venues began to drain market share from the dealers who made up the Nasdaq market. Eventually, in a bid to remain a viable market, Nasdaq sought the approval of the SEC to become a privately owned electronic market, competing with the ECNs for market share in Nasdaq securities. It has been a difficult struggle for Nasdaq. Even with the advantages of greater liquidity, Nasdaq today has less than 20 percent of the trading volume in Nasdaq securities.
Thus, by 2004 the U.S. securities market was characterized by two entirely different trading structures. For NYSE-listed securities (as well as those listed on the American Stock Exchange, or Amex), there is a centralized registered exchange that accounts for the vast majority of the trading. For Nasdaq securities, there is a market that consists of competitive trading venues, each vying to gain and hold market share. Although some portion of trading in NYSE securities takes place on Nasdaq or the ECNs under various exemptions, by and large the two different markets do not compete to offer better services or prices to investors. What competition between them occurs is limited to vying for listings by companies that want their shares to be traded on an organized market.
It is hard to see how government support for the existence of two entirely different securities market structures can have any rational policy basis. One of these trading formats must, on the whole, be better than the other. It would be one thing if the two structures were competing--if, for example, NYSE-listed securities could be traded on Nasdaq and vice versa. Then competition would resolve the issue: the best overall system would win. This is what happens when competitive and incompatible systems arise in the private sector; competition either picks a winner, or the systems become specialized so that they serve important submarkets. As an example of the latter outcome, in computer operating systems, Windows is now the dominant form, but Macintosh retains some market share because it offers special qualities for various submarkets.
But what is unusual in the heavily regulated securities market is that government regulation seems to be preventing competition, perpetuating support for two different market structures so that competition between them cannot resolve the question of which is best for investors and public companies. It is as though the Federal Communications Commission were fostering two different and incompatible telephone systems, so that users of one system could not place calls to users of the other. We would not think this made much sense if it happened in the world of telephones, but for some reason the existence of a similar outcome in the securities market has not provoked widespread objection.
Judging the Centralized NYSE Structure
Each of the structures has its strong proponents. Supporters of the centralized NYSE structure and the trade-through rule argue that concentrating most trading in a single market reduces bid-ask spreads and thus provides investors with the best prices available in the market. They also contend that breaking up this central trading venue by permitting or encouraging competition would fragment the market, so that bid-ask spreads will widen, price discovery will weaken, investors will get inferior prices, and medium- sized and small companies will have difficulty attracting investor interest in their shares. Moreover, they argue, the specialist system utilized at the NYSE--in which a single firm is charged with maintaining an orderly market in each listed security by acquiring shares when there is an imbalance of sellers and selling shares when there is an imbalance of buyers--prevents unnecessary share price volatility and the losses that can result. Such a market is said to be better for "retail" investors, whose relatively small trades usually receive what is called "price improvement" when they are bought or sold by a specialist at a price somewhat better than that currently posted. These are powerful arguments for the centralized NYSE structure.
However, many institutional investors have complaints about trading in the centralized NYSE market. Among their concerns is the view that their trading interest--either buying or selling--has a greater "market impact" when trading occurs on the NYSE than when it takes place on the ECNs. This is a particular concern of institutional investors--mutual funds, pension funds, and other large traders--that want to buy or sell large blocks of shares within a short period of time. What they mean by greater market impact at the NYSE is the effect on share prices when the existence of a large order to buy or sell becomes known in the human-mediated NYSE market; in this case, prices rise or fall as others anticipate the effect of the order and profit from trading ahead of it, adversely affecting the average price that institutional investors receive in completing a large trade.
For this reason, many institutional investors believe that they can generally get better pricing on ECNs, where they can trade anonymously and without revealing the size of their trading interest to others in the market, thus reducing the market impact of their transactions. In a 2003 survey by Greenwich Associates for Instinet (an ECN), presented at an AEI conference in October 2003, the institutional investors who participated in the survey reported that ECNs were three times more likely to deliver low market impact than an exchange and twice as likely as a negotiated broker-to-broker trade that does not take place on the NYSE floor. In the survey, ECNs were also deemed superior to exchanges in achieving anonymity, price improvement, and fast execution, but exchanges were deemed superior for certainty of execution.
As long as the trade-through rule exists, it will be difficult for institutional investors to trade NYSE-listed securities on ECNs. This is because ECNs match buy and sell orders virtually instantaneously. If, as required by the trade-through rule, these orders must be sent first to the floor of the NYSE (because a better price may be posted there), the delay before execution might mean the original match can no longer be effected. An example will make this clear. Assume that an institutional buyer wants to purchase 5,000 shares of Company A, an NYSE-listed security, that has been posted for sale on an ECN at $30. At the same time, an offer to sell 100 shares of Company A at $29.50 is posted on the floor of the NYSE. If the trade-through rule were to apply, the institutional buyer's order must first be sent to the NYSE to clear the 100-share offer before it can be executed for the full 5,000 shares on the ECN. Some surveys indicate that it takes an average of twelve seconds for the NYSE specialist to respond to an order. By that time, for a variety of reasons, the original offer at $30 may be gone. Thus, much of the value offered by ECNs for institutional trades might be lost as long as the trade-through rule continues to apply to NYSE-listed securities.
Apart from institutional investors' complaints about the centralized NYSE structure and the applicability of the trade-through rule to the trading of NYSE-listed securities, supporters of a competitive market structure like the Nasdaq market argue that the competition among trading venues encourages innovation that will provide better service to investors and reduce costs. There is some evidence for this. The ability of ECNs to take more than 80 percent of trading in Nasdaq securities away from Nasdaq itself demonstrates that innovation can produce new ways of doing things that can out-compete even established institutions. In this case, to its credit, Nasdaq preserved itself by becoming an electronic market so it could compete with the ECNs, but the ultimate beneficiaries of the new trading method were investors, including institutional investors and the many shareholders and pensioners they are trading for. If the trade-through rule had been applicable to Nasdaq, it is very likely that the ECNs would never have become a significant competitive factor in the Nasdaq market.
So what we have here is a complex policy debate, revolving around questions such as these: Will a centralized market deliver better services to investors over time than a competitive market? If the trade-through rule were eliminated, would the ECNs out-compete the NYSE as they out-competed Nasdaq? Would that be a bad thing or a good thing for investors or for companies of varying sizes that list their shares for sale? There is little doubt that ECNs are a benefit to institutional investors; the benefits they offer to individual investors are less clear. Some contend that the trade-through rule and the centralized structure of the NYSE provide greater benefits to individual investors. Should the securities market be structured so as to provide benefits to institutional investors over individual investors, or the other way around? Should this question be answered through a regulatory decision or through competition and would not allowing competition be a regulatory decision in itself?
The SEC Plan
Into this policy debate stepped the SEC in February 2004, with a market structure reform proposal titled Regulation NMS. The proposed regulation would: (1) modify the application of the so-called "trade-through" rule in the trading of both New York Stock Exchange listed securities and Nasdaq securities, (2) impose a ceiling on the market access fees used by ECNs, (3) prohibit sub-penny bids and offers, and (4) change the method for the sharing of revenue from the sale of market data.
Although all of these proposals have significance for various aspects of securities market structure, the proposed changes in the trade-through rule have raised the most serious questions about what the SEC is proposing to do.
Before the publication of Regulation NMS, many observers anticipated that the SEC would eventually have to decide whether to retain the trade-through rule or to eliminate it, and that this decision in turn would suggest whether the SEC's market reform proposal would move in the direction of encouraging a centralized market like the NYSE for all securities trading, or a competitive market--like the Nasdaq market--where trading venues compete with one another.
But when Regulation NMS was published, it turned out that it was completely ad hoc. It did not seem to be based on any vision of what would be the best market for investors or even for companies that have listed their shares. It proposed to allow investors--probably institutional investors--to opt out of the trade-through rule on a trade-by-trade basis, which suggested that the SEC favored competitive markets; but then it also proposed to apply the trade-through rule to the Nasdaq market, where it had not been applicable before.
The only plausible theory for what the SEC might have been trying to do with this internally contradictory proposal is that it was a political compromise: the agency was attempting to placate the various parties in the debate by giving institutional investors and the ECNs a limited opportunity to trade NYSE securities in the electronic markets, while telling those who favor a centralized market and the trade-through rule that all the benefits of the rule will now be available to investors in Nasdaq securities. But it does not work. If institutional investors take advantage of the opt-out provision, it could substantially reduce the role of the NYSE as a central market and thus the benefits a central market is supposed to confer. And if applying the trade-through rule to Nasdaq securities has any effect, it will destroy the benefits that many see in the competitive Nasdaq marketplace. Regulation NMS, then, may be good politics for the SEC, but it is not good regulatory policy and it is certainly not real market structure reform.
More seriously, the 200-page release that accompanied Regulation NMS did not contain any rationale for the proposal--no analysis comparing the benefits that investors or companies receive from a centralized market with the benefits they derive from a competitive market, and no suggestion of what interests and purposes the SEC believes the securities market should serve. For example, an important question is whether in fact small investors--also known as retail investors--receive greater benefits from a centralized structure such as the NYSE or a competitive structure such as the Nasdaq market. It may well be true that when retail investors place orders for NYSE shares, they get better pricing on the NYSE, but it is also probably true that most small investors are shareholders through intermediaries such as mutual funds or pension funds, and these institutions may get better pricing through use of ECNs.
This is an important issue as a matter of policy, and the SEC could have contributed to its resolution by analyzing what benefits retail investors receive directly in comparison to the benefits they receive as shareholders of mutual funds or beneficiaries of pension funds. How many "retail" investors are there, and how much trading do they do? Even if we assume that retail investors receive better prices by trading on a centralized exchange such as the NYSE, do these benefits outweigh the lower overall prices that institutional traders believe they receive by trading through ECNs? On questions such as this--which should be central to a resolution of the policy issues associated with market structure reform--the SEC provides no analysis and no answers.
Unfortunately, this follows a pattern that is becoming characteristic of the current SEC. Its controversial shareholder-access proposal, for example, also had no empirical or analytical support and seems to be based entirely on placating various constituencies. Most recently, in adopting rules requiring that all mutual funds have independent chairs, the commission pointedly ignored data demonstrating that funds with non-independent chairs perform better over time than funds chaired by independent directors.
The release that accompanied the Regulation NMS proposal noted that "the objective of market center competition can be difficult to reconcile with the objective of investor order interaction." Indeed, as we have suggested above, such a reconciliation appears to be wholly impractical. But whether the SEC tries to reconcile the two systems or, more realistically, attempts to choose between them, the task is made more difficult if the agency does not do any analysis.
Comparing Securities Markets Structures
There is a good deal of useful analysis that could be done. At AEI, we have been studying the issues associated with securities market structure since early 2003 and will eventually issue a report with recommendations for reform. In connection with this project, we commissioned Professor Kenneth Lehn, a former chief economist at the SEC and now a professor at the University of Pittsburgh's Katz Graduate School of Business, to study how the three different trading venues--the NYSE, Nasdaq, and the ECNs--respond to stress. In a paper delivered at an AEI conference on June 10, 2004, Professor Lehn and two of his colleagues, Sukesh Patro and Kuldeep Shastri, presented their analysis, which contained some important data for students of the securities markets.
One of the strongest arguments for a centralized market structure is that it promotes efficient price discovery by focusing the maximum amount of liquidity in a single place. Price discovery--finding the price at which buyers and sellers will transact--is one of the principal functions of markets, and it is generally assumed that the more liquid a market, the more efficient the price discovery process. This is logical, since the concentration of buy and sell orders in a single place creates the maximum opportunity for the right price to be found, given the information available in the market at that moment.
There are two important corollaries to this idea. The first is that traders who transact in a highly liquid market get the best possible price at any given time, and the second is that an efficient and centralized market should be less volatile under stress, because the volume of buy and sell orders should keep the market from moving too rapidly in either direction in response to favorable or unfavorable news. Thus, an investor who wishes to sell on bad news should be able to get a better price in a liquid market than an illiquid one, because the volume of buy orders in a liquid market should provide some support to the price, even though the price is moving down on adverse news.
In the specific case of the NYSE and the Amex, an additional factor is said to dampen volatility. These markets both use a specialist system, in which--as noted above--an individual (employed by a specialist firm) is responsible for assuring an orderly market by buying when there is an imbalance of sellers and selling when there is an imbalance of buyers. There is a lot of skepticism about whether specialists actually do this effectively, but studies comparing Nasdaq price volatility with NYSE price volatility have in the past shown lower volatility in prices on the NYSE.
Given these assumptions, it would be expected that Professor Lehn's analysis--which focused specifically on periods of market stress--would further confirm the superiority of the centralized market structure in delivering better price discovery and lower volatility. However, this is not what the analysis showed.
In their study, Professor Lehn and his colleagues paired 341 NYSE stocks with the same number of Nasdaq stocks traded on two days during 2003. Paired matching was necessary because, as noted above, the NYSE and Nasdaq are--with a very small number of exceptions--completely separate markets, trading an entirely different group of stocks. One of the major contributions of the Lehn study was the care with which pairs of stocks were matched, so that the overall efficiency of the different markets could be assessed for similar stocks on the same day. The match excluded all financial firms, utilities, ADRs, all firms with prices less than five dollars, and all NYSE firms with market values less than $300 million. The matched stocks were then divided into terciles (thirds) according to market capitalization. The large capitalization stocks, incidentally, account for most of the volume in their respective markets, including about 70 percent of the volume on the NYSE.
The authors then selected two days of market stress during 2003--that is, two days on which the markets were surprised by economic news, causing a rapid rise or fall in securities prices. The market stress days chosen were January 2, 2003, when the Dow Jones Industrial Average made its largest positive move that year, during forty minutes of trading in the morning, and May 1, 2003, when it made its largest negative move that year in a similar forty-minute period. In both cases, the market was surprised by reports from the Institute of Supply Management on that organization's index of manufacturing business conditions. The performance of both markets was then evaluated for the stress periods and calm periods on each of the two days chosen.
The data collected by the Lehn group is significant for policy analysis. It showed that for stocks in the largest capitalization tercile, in both calm and stress periods on those two days, both quoted and effective bid-asked spreads were lower on Nasdaq than on the NYSE, and in the Nasdaq market itself bid-ask spreads were lower on ECNs than they were at traditional market makers. On the other hand, for stocks in the two lower terciles, spreads were lower on the NYSE than they were in the Nasdaq market during both stress and calm periods. This result is in accord with other academic studies. However, in periods of stress for all terciles the spreads at ECNs widened less than on Nasdaq, and thus appeared to account for the relatively good performance of Nasdaq when compared to the NYSE on the top tercile and to improve the performance of Nasdaq for the two lower terciles, even though for these lower terciles Nasdaq's performance was not as good as that of the NYSE during periods of both calm and stress.
This result suggests that where there is already a high degree of liquidity in a stock there is no diminution in the quality of the market--including the process of price discovery and the suppression of volatility--when the stock is traded in a supposedly "fragmented" and competitive electronic market such as that for Nasdaq stocks. On the other hand, stocks with less liquidity may benefit from being listed on a centralized market such as the NYSE.
The implications of this data are profound. More research is probably necessary--the Lehn group is refining its work for presentation at another AEI conference in the fall of 2004--but as a preliminary matter, it appears that for large capitalization stocks the competitive structure associated with the Nasdaq market functions better under periods of both calm and stress than the centralized NYSE structure. The adverse effects of what is called "fragmentation"--the breaking up of liquidity so that spreads widen and price discovery suffer--is simply not evident for these securities. In fact, the opposite seems to be true: for highly liquid stocks, the fragmented market seems to be more orderly for these stocks than the centralized form.
On the other hand, to the extent that the SEC is interested in creating a better market for stocks of all companies, large and small, the Lehn group's results suggest that a centralized NYSE-style market may offer benefits for stocks with lower volume levels. This in turn suggests that if the NYSE were opened to competition by the electronic markets, trading in the high volume stocks might gravitate to Nasdaq and the ECNs, while trading in lower volume stocks would tend to concentrate on the NYSE and the Amex--where volatility and spreads would be lower during both calm and stress periods. More research is currently underway on this subject.
However, for purposes of this Financial Services Outlook report, the important point is that carefully designed and implemented studies could give the SEC the information necessary to understand the comparative benefits and deficiencies--for both investors and companies--of the centralized and the competitive structures that currently exist in the U.S. securities market. The fact that the SEC is attempting to implement market structure "reform" without doing an analysis of this kind should be a matter of concern to all those who understand the value and importance of the U.S. securities market.
1. Greenwich Associates, Instinet Proprietary Trade Execution Study: Research Results (October 2003), 8. For a link to the study and for other information on the AEI conference at which it was released, visit www.aei.org/event650.
2. To be sure, NYSE securities can be traded on ECNs without complying with the trade-through rule, but only if the ECN does not post a price in the Consolidated Quotation System--a system that lists all current posted prices for NYSE and other listed securities. However, without the opportunity to advertise a price, it is extremely difficult for ECNs to attract trading interest in NYSE and other listed securities.
Peter J. Wallison is a resident fellow at AEI.