March 28, 2000

Jonathan G. Katz
Secretary
U.S. Securities & Exchange Commission
450 Fifth Street, NW
Washington, DC 20549-0609

Re: Release No. 34-42450; File No. SR-NYSE-99-48; Market Fragmentation

Release No. 34-42208; File No. S7-28-99; Regulation of Market Information Fees

Dear Mr. Katz:

I am Steve Wunsch, president of AZX, Inc., operator of the Arizona Stock Exchange. I welcome this opportunity to comment on two recent Commission releases seeking public views: that on market fragmentation, and that on market information fees and revenues. As both releases make clear, the two issues are closely related. Accordingly, I submit this letter in response to both requests for comment. Further clarification is available at www.azx.com;(see Auction Countdown Commentaries, particularly Price Discovery Error (3/9/98), Mayday Mayday (5/1/99), and Branding Sand (10/1/99), enclosed).

My view, in sum, is that the market design interventions undertaken under the National Market System (NMS) rubric are fundamentally flawed policies. They are based on incorrect market structure theory in such key areas as transparency and fragmentation, and the authority to implement them is based on flawed competition doctrine, namely, antitrust. In view of the importance of capital markets to our economy and society, I would urge the Commission and the Justice Department to review the appropriateness of applying antitrust to stock markets. Following current policies to their logical conclusion - a CLOB (consolidated limit order book), or similar arrangement - is likely to produce severe volatility and a crippling misallocation of capital.

As you know, AZX runs single price call auctions in exchange-listed and OTC securities ending at certain fixed times during the day, pursuant to a low volume exemption from exchange registration. What you may not know is that AZX's call auctions had their roots in attempts in the `Eighties to bring transparency to stock and futures markets, so as to improve centralization, price discovery, stability, and liquidity. Since the instant releases are laced with these and related issues, I would beg your indulgence to consider the following short history of AZX's roots, as it bears on both the issues themselves and on AZX's interest in them.

Sunshine Trading

For most of the `Eighties I worked in the Financial Futures Department at Kidder Peabody, helping our institutional customers devise strategies to reduce trading costs, particularly for large trades. This became especially important for managers of "portfolio insurance" strategies, because they needed to buy increasing positions in rising markets and sell them in falling markets. As their strategy gained acceptance and implementation authority, primarily from pension funds seeking to protect their assets, it became harder and harder to squeeze their trades into the market without intolerable market impact. At the extreme, their strategy would collapse if their buying were causing the market rises to which they were responding with more buying. More poignantly on the downside, some were asking the question: Could minor downward movement trigger sales that would trigger more downward movement and further sales in a descending spiral, or "meltdown?"

To address this risk, we at Kidder developed a strategy called "sunshine trading,"¹ in which an investment manager would announce something like the following to the market: "At 11:00 we will be offering 2,000 S&P 500 futures contracts at roughly the market midpoint at the time. If all contracts are not sold within a minute, we will cease offering, but will return every 15 minutes with the same procedure until all 2,000 contracts are sold."

Facilities to accommodate such pre-announcements of trading intent were (and still are) non-existent in futures markets. More problematic still was the fact that futures market rules had difficulty distinguishing between honest efforts authorized by customers to improve liquidity by pre-announcing, and "pre-arranged" trades, which are prohibited, allegedly in order to protect customers. While some exchanges, notably the New York Futures Exchange and the Chicago Mercantile Exchange, tried to redress such shortcomings, their proposed rules lost momentum amid the general din of concern over escalating volatility, which many blamed on "program trading."

At a hearing held by the House Subcommittee on Telecommunications and Finance on July 23, 1987, concerns over program trading were expressed and rebutted by 15 witnesses, including the heads of the major exchanges and the CFTC. My own testimony recommended that "[t]he CFTC should clearly establish that sunshine trading is permissible and encourage futures exchanges to adopt procedures providing for the voluntary dissemination of accurate trading information." I also said authorities should "[e]ncourage the development of efficient and competitive single price auctions at the open, the close and once intraday for stock exchanges, futures exchanges, options exchanges and OTC markets."

I recount this history to establish the link between transparency and single price call auctions. On a micro level, placing an order on AZX's open book accomplishes the pre-announcement of an intent to trade, just as the sunshine trading strategy cited above did. The purpose was and is to centralize counter-party interest by focussing everyone on trying to trade at a fixed time known in advance by everybody, thereby allowing their liquidity demands to naturally offset each other. And the sunshine pre-announcement (the AZX open book order) disseminates far more valuable information than traditional continuous order or quote screens, because the viewer of the sunshine order book has time to respond and participate in the trade, and because the order book aggregates many such orders for simultaneous execution.

On a macro level, the temporal centralization of trading at the moment the call ends (the sunshine pre-announcement's time of execution) dramatically improves the quality of price discovery information by aggregating supply and demand at that time. In other words, not only is such temporal centralization important for liquidity, but it is also the best way to balance supply and demand in the market, thereby mitigating volatility and, in stressful times, averting or stopping a crash. Importantly, the micro and macro interests are not in conflict. Given the proper facilities to accomplish sunshine pre-announcements, the interest of traders in liquidity is consistent with the interest of the broad market in stable and accurate price discovery. There is no need to compel participants "in the public interest" to do what they do not perceive to be in their own interest.

As my 1987 testimony described at length, the opposite of a single price auction's centralization and stabilization is the temporal fragmentation of continuous trading, which causes "excess volatility," "imposes an entirely unnecessary cost on users of the market," and risks "cataclysm." My discussion of the wild volatility on January 23, 1987 - one of the main reasons the hearings were called - made clear my belief that the continuous market was the principal problem.² My warnings and recommendations at that pre-Crash hearing (the market was just about to peak) went unheeded. Mine was but one of many voices in a cacophony of bull market euphoria and excitement over new investment technologies, strategies, and products. Apparently, the increasingly frequent and strange volatility events seemed but a small price to pay under the circumstances.

Today's Dangers

The dangers I saw in `87 are far more prevalent now, primarily because our markets are far more continuous now. Not only are there no meaningful single price auctions yet, but the continuous market itself is much more continuous. And not only are there still no facilities encouraging the voluntary release of accurate information, but mandatory transparency has been steadily working against its own goals. The recent ratcheting up of mandatory transparency via the antitrust settlements, the order exposure rules, and narrower increments are, in my view, the primary cause of the dramatic escalation of volatility. These measures are working against transparency by forcing large participants to hide their size and information even more, with big blocks once handled discreetly at accurate prices upstairs now increasingly broken into hundreds or thousands of tiny pieces, machine-gunning the market with momentum swings. One institutional trader told me recently, for example, of executing a 1.4-million share order in ten minutes at an average trade execution size of 2,300 shares - a little over a trade per second from that one institutional order alone. This works against transparency, first by preventing the more accurate (block) price from forming, and then by spewing the incorrect and rapidly changing prices at such blinding speed into the market that they are little short of incomprehensible. The resulting price discovery distortions are probably responsible not only for the much-noted increase in volatility, but also for the increasingly grotesque valuations. And they are allowing bubbles to form, facilitating fraud, and raising the danger of crashes with their attendant systemic risk.

These problems are not the result of random exogenous factors, such as the advent of new technologies, or the excessive exuberance of online traders. They are the unintended consequences of the policies on which the releases seek comment, namely, those meant to bring transparency and centralization to the trading process. While antitrust law certainly supports separating the old membership markets from their information monopolies, and while much academic opinion certainly supports linking regional exchanges (and ECNs) in a "national" market system, these well intended policies are in reality dismantling price discovery and thereby destroying the market's ability to adjudicate value. As discussed below, the loss of effective price discovery is resulting in the greatest - albeit inadvertent - short squeeze in history. New shares are rising primarily because most people don't own them yet, and old shares are falling primarily because people do own them. The value - relative or absolute - of their underlying businesses is becoming increasingly irrelevant. And the volatility of the momentum see-saw between Old Economy and New Economy stocks, as the short squeeze plays out or threatens collapse, plays into the hands of chat room scamsters who fan the flames of sudden fads that are just as quickly reversed.³ The inevitable consequence will be a massive misallocation of capital - away from productive and promising enterprise and into businesses with far less potential than investors will have paid for them.

The Missing Dimension

The fundamental theoretical error driving these mistaken approaches toward transparency and fragmentation is a failure to recognize that fragmentation exists potentially in two dimensions - not just one; that fragmentation can exist in the temporal dimension, as well as in the obvious spatial dimension. Tying together disparate places, be they regional exchanges or ECNs, has provided the entire rationale for NMS, including all transparency policies, all linkage policies, all best execution rules, the order handling and exposure rules, even decimals.⁴ The instant release on fragmentation, for example, poses several possible degrees of mandatory price- and time-priority rules and methods to implement them. Such measures are purportedly needed in order to make sure that an order receives the best price available, but only at a particular point in time. No concern has ever been expressed, either in this release or in any other regulatory forum of which I am aware, that buyers and sellers could just as easily miss each other by arriving at the market at different times. And no concern on fragmentation grounds has ever been expressed over a buyer who pays a suddenly high price that is just as suddenly reversed. Such an oversight could perhaps be dismissed as a merely semantic technicality, were it not for the fact that spatial fragmentation provides the implicit justification for the entire NMS effort. But - if temporal fragmentation were properly taken into account - it would become clear that total fragmentation, net of both spatial and temporal varieties, is bound to rise as a result of the NMS effort to reign in spatial fragmentation. In other words, as discussed below, if temporal fragmentation were taken into account, NMS would be seen clearly as harmful to its own goals.

Ironically, both types of fragmentation were addressed reasonably well by the membership model of exchange organization, which is headed for extinction under NMS. Spatial fragmentation was resolving naturally as single exchanges won effective monopolies in the trading of their listed stocks. Temporal fragmentation was addressed by upstairs trading in which large institutions privately revealed to large dealers the needed information for those dealers to service their liquidity needs. Importantly, both capital-preservation and reputation considerations called for honest dealing and were mainstays of accurate price discovery. It is ironic that the membership model of exchange organization was so effective at addressing fragmentation - the raison d'être of NMS - because the membership model is under such sharp attack by NMS reforms that it is disintegrating. That naturally centralizing membership exchanges are planning to "demutualize" suggests that NMS - however well intended or legally justified - is creating the problem it was designed to solve.

Granted, the exchanges were rife with antitrust violations. Indeed, it could be fairly said that the main purpose of exchange organization, from the members' viewpoint, was to gain privileges unavailable to the public. Almost all of these privileges are probably illegal under a strict antitrust interpretation. They included the abilities to have trading information the public did not have and to fix commissions, spreads, minimum price increments, and many other trading rules that favored their members over the public. They also included deliberate attempts to monopolize trading in their stocks by freezing out competitor exchanges with off-board trading restrictions. All of these probable antitrust violations form a clear and compelling legal basis for NMS reforms. Yet I would urge the Commission to conduct a review - perhaps in conjunction with the Justice Department's Antitrust Division - of the founding documents and understandings of the world's major capital markets. Could any of the major exchanges have formed if antitrust principles were being adhered to at their founding? I suspect not. If not, then the Commission's NMS reforms could be seen as springing from a perceived need to have Government redesign the most important structures and functions of our capital markets. The complexities and difficulties of such a task are immense, and, in my view, beyond the competence of any agency but a free market unhindered by interventions like NMS.⁵

Disintegration Could Accelerate

In any case, as the exchanges demutualize themselves away from the structure that gave rise to both antitrust violations and natural solutions to fragmentation, the potential dangers of continuing to impose NMS's linkage-based solutions are considerable. Chief among these is the danger that any gains made addressing spatial fragmentation will increase temporal fragmentation. It is probable that, in the absence of membership organizations, fragmentation is natural and inevitable in continuous markets, and, like a balloon, squeezing it down in one dimension will only cause it to pop up in the other. But, from a public policy perspective, spatial fragmentation is much less damaging than temporal fragmentation. The fairness issues driving NMS's focus on spatial fragmentation, while seemingly important, actually pale in comparison to the potential damage done by temporal fragmentation, which can include excessive volatility, valuation distortions, systemic risk, and an easier path to committing various price-manipulation frauds.

A CLOB, for example, may sound like a clean way to solve, once and for all, the fragmentation problems posed by regional exchanges and ECNs. Indeed, a CLOB is the logical endpoint for all NMS efforts since inception - the ultimate "national" exchange.⁶ But there are many questions to consider. First, the Commission should be aware of the powerful inducements to counterproductive rent-seeking behavior its involvement in such matters creates, irrespective of whether or not it decides wisely. Many of today's major participants will be greatly affected by how tightly the Commission turns the time-priority dial, from standing pat with today's loose linkage, to creating nationwide price- and time-priority rules, to mandating that all trading occur in a single system with price- and time-priority. Just as the order handling rules gave life to the new ECNs (primarily at Instinet's expense) by enabling them to free ride off public liquidity, the Commission's fragmentation choice could create whole new businesses or cause others to go up in smoke. Judging by their vehement public arguments, strict time-priority will harm the businesses of certain exchanges, ECNs, and e-brokers, just as much as it will help - for that reason - certain others. Whether these competitors are right or not in their analyses, the Commission's decisions are increasingly critical to the viability of many businesses. This is a weighty responsibility, not to be undertaken lightly or without great confidence in the efficacy of the policies in question.

Such confidence does not appear to be in hand. The absence to date of any recognition that temporal fragmentation even exists suggests that the Commission's fragmentation policies are on shaky theoretical ground. Since temporal fragmentation was not considered in the fragmentation release, none of the discussions therein provide a sufficient basis for undertaking any of the choices outlined. If the arguments offered here are deemed by the Commission to have any chance of having merit, then a thorough re-evaluation of the entire NMS policy is advisable. As I see it, not only are the most uncomfortable features of today's market, such as increased volatility, increased fraud, and valuation distortions, at least partly attributable to the policy, but upping the ante with further and tighter linkage could be very dangerous.

The closer NMS gets to CLOB, or system-wide price- and time-priority, the more volatile and unstable markets are likely to become, primarily because CLOB will magnify the size-shrinking effects of all transparency policies. It appears that NMS policies have already contributed to making the markets considerably more continuous in recent years - indeed, "hyper-continuous" is not too strong a word to describe today's rapid-fire trading. This contributes to volatility by breaking down price discovery into such small pieces that there is little chance to find a consensus value for supply and demand. The less each trade represents of total volume, the more uncertain and unstable price discovery becomes, and the more volatile the market. It is even probable that the market's newfound tendency since the mid-`Eighties to "crash" from time to time has been aided and abetted by NMS policies as old as the tape and ITS (Inter-market Trading System).⁷ And it is likely that recent NMS initiatives, such as the order handling rules and narrower increments, are further augmenting continuousness, as will decimals when they are introduced. The average trade size as a percent of total trading volume, the inverse of which is perhaps the best gauge of continuousness, has already been dropping fast under recent reforms.⁸ Since narrowing increments from eighths to sixteenths was probably one of the chief causes of today's increased continuousness, moving to pennies will almost certainly be far worse, causing great volatility, instability and valuation distortions.

Temporal fragmentation and hyper-continuousness would be beneficial if the conventional wisdom on price discovery were correct, namely, that every price is a correct representation of supply and demand, whether created by a once-a-day Walrasian tatonnement (a form of single price auction), or a constantly moving sub-second stream of tiny prints. Indeed, if this conventional view were correct, the more NMS (or any other source of temporal fragmentation) broke trading down into smaller pieces, the more efficient the markets would become, because of the more regular and continuous information contained in the more frequent prints. The accuracy of the information is a given in this view, which most people share, including some very prominent and savvy people. Federal Reserve Chairman Alan Greenspan, for example, appears to attribute the strange valuations in today's market to irrational behavior, a sort of collective dementia, for which the remedy is higher interest rates and an economic slowdown. If the prices in the market correctly reflect investors' collective appetite for stocks, this conclusion may be quite logical.

But, while not discounting the possibility that incorrect prices, however produced, could cause irrational behavior, the current hyper-continuous NMS market is likely to produce incorrect prices in any case - and not by just eighths and quarters, but by massive amounts, and for indefinite periods of time. At best, this complicates the Fed's task of managing inflation. For example, in an environment of rudderless valuations in a hyper-continuous market, higher rates may eventually bring down the economy, but in the meantime they can only feed the bubble by forcing investments out of rate-sensitive Old Economy stocks, with nowhere to go but to New Economy stocks. As for the hope that slowing the economy and the wealth effect will cause a return to rational pricing, that is unlikely. Under NMS, there is no guarantee that stock prices will ever reflect economic reality - before, during or after any rate-induced slowdown. They could remain just as stubbornly undervalued at the depths of a depression as they are overvalued now. And any new capital could just as easily flow to companies with no business plans or prospects and away from solid businesses as it does today.

The Great Short Squeeze: A Flight From Quality

The technology sector's valuations are perplexing almost everyone today. One possible explanation would focus on the fact that its shares are new and scarce, and the newer and scarcer they are, the more their valuations seem able to rise to perplexing levels. It would be troubling if scarcity did have an influence on pricing. While many items, from currencies to precious metals to beanie babies, have scarcity value, companies whose shares trade on the world's stock markets are assumed to be priced based on the potential earnings of their businesses, not the rarity of their shares. Indeed, the principal social value of stock exchanges rests on their capacity to provide after-market liquidity in widely held public shares. That liquidity value is based on non-scarcity. But, with price discovery impaired by hyper-continuousness, flow-driven momentum can carry prices far away from value, and there is no arbitrage or value-based trading source that can force them back to reality. In this circumstance, scarcity could, indeed, be a significant factor driving price. If so, features that would ordinarily make shares more valuable may actually work against them. Widely held shares in companies of recognized value may tend to fall because of those factors. And illiquidity - or even a lack of earnings - may actually play in a new stock's favor, because value-based investors would tend to shun such shares. Why would their shunning the shares make them more valuable? Because, the more they are shunned, the more pent-up demand and flow that will be unleashed eventually, if these companies - by hook or crook - can maintain their capitalization toehold. That is because stocks that become part of the landscape of possible holdings are impossible to ignore when other investors hold them, regardless of the companies' prospects. The inducement to buy becomes irresistible and, indeed, automatic for many, when stocks go into the popular indexes.

Perhaps this explains why the natural scarcity due to newness appears to be deliberately enhanced by bankers and issuers who only dribble out small amounts in public offerings, so there is no way for investors in aggregate to own them in proportion to their market caps. This sets them up to rise regardless of value until value investors finally capitulate and buy them when they go into the indexes or otherwise become accepted as possible holdings. Institutional investors, such as mutual funds, like to think of themselves as making each purchase or sale decision on its own merits. But, because of the way their performance is measured versus the popular indexes, the competitive dynamics of their businesses require them, in aggregate, to own stocks in rough proportions to their market capitalizations. That is to say that, in aggregate, the mutual fund industry behaves as if it were an index fund. Call them the "have-to-buys," because they must, in aggregate, own in market proportions all new stocks that are created. In this scenario, old company shares could be falling simply because they must be sold to make room for the new ones. New stocks that can attain capitalization greater than their representation in the portfolios of the have-to-buys must be bought regardless of value. Similarly, stocks that are held in greater than market cap proportions will have to be sold - regardless of value - when the new ones are bought. Perversely, then, the more a new stock appears to be overpriced, the more buoyancy it may have in the market. By causing value investors to hold off buying, the more eventual demand and capitulation flow that will be created when value investors throw in the towel. The situation amounts to an accidental short squeeze.⁹

And here's the clincher: the higher the market caps of the new shares before value investors capitulate, the more of them that will have to be bought, and the more of the old shares that will have to be sold. In other words, as with portfolio insurance in '87, there is a powerful and very dangerous positive feedback loop at work here. The higher the New Economy's market cap relative to the have-to-buys' aggregate holdings, the more its stock must be bought. The lower the Old Economy's proportionate market cap falls below the average fund's actual holdings, the more its shares must be sold. As with portfolio insurance, mutual funds depend for their viability on the market being able to find correct prices irrespective of their trading maneuvers. But it cannot. As in the Crash of '87, when portfolio insurance sales were causing the declines that were triggering more sales under their strategy, the presence of mutual funds' have-to-buy demand for new stocks is causing many new stocks with increasingly slim plans for profits to be issued - and to rise. And the more they rise, the more new ones are issued, which also must be bought. More dangerous on the downside, the fact that funds' performance-based competition exists is forcing them to sell the old stocks, regardless of value, because they are over-weighted in them relative to the new ones. Like portfolio insurers in the `Eighties, all the have-to-buys would be fine now - if the conventional view of price discovery were correct. If correct prices were always discovered, even required trades would not cause these strategies to self-destruct. But, because of the lack of accurate price discovery in the new hyper-continuous market, their trading needs may contain the seeds of their undoing.

The collapse of the market under the weight of portfolio insurance in '87 seemed devastating at the time, but of course the market recovered. And, happily, since most of the investors using the strategy were sponsors of defined benefit pension funds, the individual beneficiaries of those funds were not harmed. But the threat now is much broader. In spite of the growing numbers of individual do-it-yourself investors created by recent NMS reforms, a great many assets of individuals are still held in mutual funds holding the old stocks. If the short squeeze described above guts the value of those stocks, many individuals' financial security will be severely harmed. And very few of them will be able to successfully extricate themselves from that calamity by getting in and out of the new stocks before most of them crash, too.

As sad as such an outcome would be, there is an even greater danger: resource misallocation. In a trenchant analysis of the current situation contained in a March 27, 2000 New Yorker magazine review of Yale economist Robert Shiller's new book Irrational Exuberance, John Cassidy says:

The investors' tendency to run in packs is encouraged, of course, by the fact that, as long as they stick together and avoid the temptation to sell, they are often rewarded by higher prices. Such a "positive feedback" effect is present in almost all self-reinforcing processes. Shiller skates over this idea, but it is central to the stock-market boom. He cites a poll published by Barron's in May of last year, which found that seventy-two percent of money managers believe the stock market is in a bubble. This is a truly remarkable finding. It suggests that the vast majority of professional investors are well aware that current stock prices make no sense, but they keep buying anyway. Why would they do such a thing? Because of the way in which they are paid. In most cases, a fund manager's performance (and remuneration) is measured not in absolute terms but relative to the performance of other people who manage similar funds. If he (or she) turns bearish and everybody else remains bullish, he is likely to be fired for underperforming. If, on the other hand, he stays bullish and the market falls, he will be in the same sinking boat as everybody else, and he will probably survive. In this environment, the key to success is to guess what everybody else is going to do next - dump blue chips, buy technology stocks - and nip in ahead of them.

Later in the article Cassidy quotes Shiller from an interview, bringing out the capital misallocation risk of such non-value-based momentum trading:

"The Internet craze is causing distortions," he said. "People are starting up little dot-com companies even though they don't personally think they're going to work. They don't care. They're just looking to sell them. This process is obviously wasteful." . . . [T]he Internet may be diverting valuable resources, human and physical, to worthless pursuits.

Regulation of Market Information Fees and Revenues

The release on regulating market information fees and revenues is commendably frank in its description of the current structure as resembling a regulated public utility. And there is no question that the Commission has correctly concluded that trading information was being kept from the public under the old exchange arrangements, which were anti-competitive under antitrust law. Yet I would urge the Commission and the Justice Department to review the role that information advantages played in exchange formation. If it turns out, as I suspect, that members' anticipated advantages over the public with respect to trading information constituted pivotal inducements to their banding together as exchanges, then far more caution is advisable in this area than has been evident in the implementation of NMS to date. While the release makes clear how difficult it is becoming to balance everyone's interests fairly as the Commission seeks to manage the market's information function as a utility, there are far bigger issues at stake. For example, what is the long run effect of having wrested control of the information function from private hands? Will exchanges still form? (The demutualization trend is not heartening in this regard.) Will increasing the quantity of prints decrease the quality of price discovery? Without answers to such questions, there is little likelihood that transparency theory could be considered sufficiently advanced to justify upending traditional information structures.

The Commission should also review transparency theory, itself. It is likely, in my view, that the harsh glare of the tape is responsible in significant measure for the breakdown in trade size that is causing the volatility, instability, and valuation distortions described above. A useful question to ask in any review would be: If continuous electronic transparency enhances price discovery - or at least does not harm it - why not provide a running tape of voting totals on election day, too, so voters could see how others voted before making their own decision? We do not allow such transparency in voting, of course, because doing so would introduce distortions and unfairness. A review of such questions will reveal that the Commission's transparency policy is making the market vulnerable to price discovery errors and inviting manipulation, which lead in turn to all of the problems described above.

Conclusion

There is no question that NMS reforms are an arguably correct application of antitrust law. But, in view of the wrenching changes they are causing, an urgent review is advisable. In particular, an opinion should be formed on whether temporal fragmentation is a factor that should be considered. Is it a good thing or a bad thing that trading is breaking down into smaller pieces, as shown in the following appendix? If it is beneficial, such as by creating more prints and information in the market, then full speed ahead. If it is harmful, such as if NMS policies are causing distortions and inaccurate prices to be disseminated, then a thorough reconsideration of NMS and, indeed, of antitrust law, is in order. The litmus test is this: Does the Commission believe that all prints are accurate reflections of supply and demand? If not, then it would be appropriate to reconsider the wisdom of disseminating them through a mandatory tape. The Commission should also ask whether that mandatory tape - and the full panoply of related NMS policies - are contributing to any pricing inaccuracies.

Please do not hesitate to call, should you have any questions, or wish to discuss these matters further.

Sincerely yours,

Steve Wunsch
President

Appendix:

Average Trade Size as a Percent of Total Volume

	1993	1994	1995	1996	1997	1998	1999
AMZN					0.2779%	0.0089%	0.0049%
AOL				0.0604%	0.1100%	0.0113%	0.0072%
C	0.1165%	0.1000%	0.1459%	0.0740%	0.0663%	0.0597%	0.0221%
CMGI		12.8205%	3.0303%	3.1447%	2.0747%	0.0355%	0.0087%
CPQ	0.2630%	0.0738%	0.0860%	0.0998%	0.0269%	0.0094%	0.0166%
CSCO	0.0715%	0.0445%	0.0291%	0.0203%	0.0129%	0.0019%	0.0044%
DELL	0.3492%	0.1107%	0.0548%	0.0279%	0.0165%	0.0016%	0.0033%
GE	0.1512%	0.0606%	0.0809%	0.0643%	0.0332%	0.0119%	0.0234%
IBM	0.0720%	0.0870%	0.0850%	0.0813%	0.0488%	0.0342%	0.0148%
INTL	6.3291%	3.7594%	0.8913%	1.0183%	0.4039%	0.7716%	0.4460%
LU				0.0347%	0.0461%	0.0074%	0.0130%
MO	0.0625%	0.1123%	0.1119%	0.0781%	0.0464%	0.0326%	0.0447%
MSFT	0.0705%	0.0519%	0.0230%	0.0282%	0.0153%	0.0043%	0.0044%
ORCL	0.1321%	0.0805%	0.0525%	0.0345%	0.0212%	0.0138%	0.0106%
PFE	0.1784%	0.2189%	0.1110%	0.1369%	0.0476%	0.0197%	0.0187%
QCOM	0.2342%	0.1270%	0.0766%	0.0890%	0.0559%	0.0593%	0.0097%
WMT	0.0534%	0.0565%	0.0685%	0.0938%	0.1071%	0.0304%	0.0177%
YHOO				0.5192%	0.0712%	0.0030%	0.0046%

Source: Ozenbas, Deniz and Schwartz, Robert A., Intertemporal Fragmentation of Trading, Baruch College working paper in process.

Footnotes

¹ "Sunshine trading" is the term suggested by John O'Brien, co-founder of Leland O'Brien Rubinstein, the original and largest firm offering dynamic hedging, or portfolio insurance strategies. Other terms we had been using prior to O'Brien's suggestion included "pre-announced trading," "open packages" (for pre-announced program trades), and similar variants.

² My testimony read, in part, "This danger [of cataclysm] is all the more real in view of the fact that, in spite of what the newspapers said, January 23 was not a day on which there was much program trading activity, either of the arbitrage or portfolio insurance varieties. Neither was there any particular news to cause the morning buying and afternoon selling. I would surmise, therefore, that the probable cause of unnecessary volatility on January 23, as on any other day, was the naturally discontinuous arrival of buys and sells. To the degree in which facilities could have been provided allowing these buys and sells to arrive simultaneously, but were not, there was an unnecessary accessing of the market's risk-taking facilities, which not only cost the buyers and sellers who could have arrived simultaneously an unnecessary market impact cost, but the volatility associated with that market impact scared the daylights out of institutions and individuals alike."

³ The impression of casino, rather than serious stock market, is conveyed in the Thursday's Market column of The Wall Street Journal on March 17, 2000: "Never in recent years have investors seen such sudden, day-to-day shifts in what sector of the stock market is up and down. The phrase `Internet time' has become the accepted shorthand for the ability of the stock market to telescope into a morning's trading events that once took a month to play out."

⁴ Decimals were justified largely on transparency grounds, demonstrating the inevitable interconnectedness of all NMS policies. And, according to the March, 2000 issue of Institutional Investor magazine: "Wall Street officials first began talking with legislators about moving to decimals as far back as 1975, as part of the larger effort to create a "national market system" of linked exchanges and over-the-counter dealers."

⁵ A policy review is needed for reasons that go beyond the application of antitrust to stock markets. With a variety of Internet B2B exchanges forming now, regulators are hastening to apply competition policy to them. (See the March 13, 2000 Industry Standard magazine's "Regulating Exchanges" article). Absent a review, the default policy will be that by which NMS is reforming such powerful markets as the NYSE, Nasdaq, and Instinet, rather than the almost diametrically opposite policy under which they formed. Do we want the Internet, sometimes described as this century's railroads, regulated the way railroads are regulated now, or the way they were regulated - or not regulated - when they were built?

⁶ The National Market System was originally called the Central Market System, and its key component was a consolidation of all markets in a single electronic book, or CLOB. Many academics, then and now, consider NMS and CLOB to be virtually synonymous, as do significant numbers of regulators.

⁷ That the Crash of '87, an 18 standard deviation event, could occur without a consensus explanation for its cause(s) is remarkable. One possible contributor is that NMS policies, such as increased transparency via the transaction tape and ITS linkage, had undermined price discovery sufficiently by then to create an unstable environment. This unstable environment may have enabled other factors - portfolio insurance, program trading, mutual fund liquidations, etc. - to wreak havoc.

⁸ See Appendix for preliminary data on the breakdown of trade size.

⁹ There is some precedent for an accidental short squeeze. In 1901 - another age of volume records and frothy speculation - a battle between J.P. Morgan and a rival faction over control of Northern Pacific Railway pushed its price from under 100 to 146 before Morgan reasserted control. But many speculators, seeing the obviously unjustifiable prices, and unaware of the battle for control, shorted the stock during its rise to 149 ¾ from 127 ½ on May 6. Then, although neither Morgan nor his rivals were trying to catch the shorts in a squeeze (indeed, Morgan attempted to make shares available to them at 150) the squeeze was on. According to the account in Jean Strouse's Morgan, American Financier, "[o]n May 7 and 8, 1901, other stock prices crashed as the NP shorts dumped everything they had to cover their sales, and on `Blue Thursday,' May 9, NP leaped to a preposterous $1,000 a share. Speculators grimly realized the market was effectively cornered: they had sold 100,000 more shares than had ever been printed, and could not buy the stock at any price."