 |
Home | Previous Page | |
|
||
 |
|
|
|
||
 |
|
|
|
|
|||||||||||||||
|
|
|||||||||||||||
|
|
Division of Market Regulation:
| ||||||||||||||||||||||||||||||||||||||||||||||||
| I. | Report by Don Langevoort: General Description of Alternative Model | |
| II. | Technology | |
| A. | Don Langevoort: What Technological Standards Need to be Set? | |
| B. | Tom Demchak: SIAC Presentation | |
| C. | Mike Atkin: Summary of Vender Survey | |
| D. | Discussion | |
| III. | Economic Model | |
| A. | Annette Nazareth: Summary of Display Rule and Best Execution Obligations | |
| B. | Don Langevoort: Summary of Potential Benefits and Costs | |
| C. | Simon Johnson: Economic Issues | |
| D. | Discussion: How a Multiple Consolidator Model Would Work with the Display Rule Should the Display Rule Be Rescinded? Would the SEC's Role Need to Change? If so, how? | |
| I. | If the Display Rule is Retained, How Should Information Not Subject to the Display Rule (i.e., Deeper Data) Be Treated? | |
| A. | What Should Be the SEC's Standard for Reviewing Fees (If Any)? | |
| B. | Should Fee Filings Be Made with the SEC? | |
| C. | Should Market Makers and ECNs Be Permitted To Sell Their Data Directly to Vendors and Others? | |
| II. | Other Open Issues | |
Mr. Seligman: Let me call our meeting to order. This is the fifth meeting of the SEC Advisory Committee on Market Information. During this meeting we anticipate completing our review of equities and market information prior to the preparation and initial circulation of one or more draft reports, and I anticipate there will be more than one.
The focus of our meeting today will be on what we have referred to as multiple consolidators, and because the technological and economic issues here were particularly complex, I asked one of our committee members, Professor Don Langevoort, to chair a subcommittee to prepare the agenda for today.
The subcommittee was open to any member of this committee who wished to serve on it, and 11 members did serve. We made it a point, consistent with the Federal Advisory Act, that the subcommittee met off the record. That is, there were no transcripts prepared. Unlike everything else in the process or virtually everything else, there was a private discussion. The purpose was somewhat like an attorney work product to help prepare the discussion today, which is the official discussion on the issue.
Anyone who was a member of that subcommittee has the same right to submit any document he or she would wish for our record, and as you will see, Don Langevoort, Simon Johnson, Mike Atkin prepared various documents that have been submitted to the full committee and will be part of our record.
When I say "part of our record," I both mean the official record kept at the SEC and because we live in an age of technology, these documents will also be available through our website or through the SEC's website on the subject, as well, of course, as meeting agendas, meeting minutes, and so forth, and meeting transcripts.
Now today, let me highlight just a couple of other points. At the conclusion of our meeting I want to walk around the table and give everyone an opportunity to make final observations or remarks with respect to equities issues. This can be of enormous value to me and others who are working on the initial draft report. We are very eager not to miss anything. We are very eager to capture all of the nuances and points of view of those on the committee.
The draft report, as I mentioned, will be circulated I anticipate at some point early in June in its first version, and we'll keep going through round-robin circulation till we are comfortable we have it factually accurate and we have accurately captured everyone's views.
I don't expect that there will be total unanimity on every aspect of the report, although I am willing to be surprised, but I will make it a point to ensure that differences are effectively reflected in the report. This can either be in the text, this can be in footnotes, or in separate statements that will accompany the ultimate report.
Finally, let me remind everyone that the last meeting of the Advisory Committee will be on July 19th and my intent is that that meeting will focus on options and particularly how whatever our reports suggest would be appropriate for equities might be applied to the options side. We are all well aware there are significant differences in market information in the options context than in the equity context.
I do not anticipate forming a subcommittee in a formal sense to prepare for the options meeting, but I will have a conference call probably early in July with any who wish to participate in it, and the purpose will be somewhat like the subcommittee that Professor Langevoort chaired to focus on structuring the best agenda for our July 19th meeting.
Under the Federal Advisory Act, all of our official action must be in open public meetings and I will remind those in the audience that they will have the opportunity at the conclusion today to ask questions or make comments and any decision that is made must be by the full committee rather than by any subcommittee.
With that said, let me ask Don Langevoort to give a report on his subcommittee and then we are going to focus this morning on what emerged as the two pivotal issues, Technology and the Economic Model.
Don has expressed to me a belief that in terms of weighing the significance of the issues, he would like the technology discussion to be completed, if possible, by about 10:30 and then to leave considerable time for development of the economic model. Don?
Prof. Langevoort: Thank you, Joel, and I do want to thank the members of the subcommittee for their participation in this extracurricular activity.
We did have, I think, a very fruitful discussion. Members of the full committee can judge whether it was productive, but I think we did more forward the discussion of what the world would look like under an alternative model of data consolidation.
I am going to try to be very brief. As Joel said, there are lots of issues to discuss. So what I am going to do is simply outline for you what the model looks like that we were working from and the move on to the way we saw the policy issues falling out.
When I say the model we were focusing on, our charge was to consider all alternatives to the current system, and we had two meetings in which to do that. What we did was simply choose what we saw as the most logical mechanism, which is to introduce a strong dose of competition into the process by which data is consolidated and distributed.
The vision we were working from was one in which there is no mandatory consortia. Each market center is free to sell its data on the terms that it could negotiate to a set of competing consolidators.
We had some discussion, as you will see, in terms of predicting how many consolidators there might be. We are not necessarily assuming that there will be many or assuming that there would be more than one.
The key in what I said is each market would be free to negotiate with one or more consolidators. Markets could choose, and one suspects that some of the secondary, nonprimary markets, would join in consortia that would work together on this, but certainly the primary markets, New York Stock Exchange, Nasdaq perhaps, others, would choose to go it alone and sell their data under terms and conditions that they set without going through the complicated governance process that we see today.
Consolidators would then take the data and package it in conformity with whatever rules and regulations prevail with respect to the consolidation process. Now I emphasize that because very early on in our discussion and indeed in some of the meetings of this full committee that were held in the fall the issue was put on the table as to whether the world of vigorous competition in the dissemination of market data should retain the display rule. That is an open question. We made no judgment.
But it is an important question, because in a world in which there is a display rule, what consolidators have to do, then the nature of the market structure is different from what that world would look like without the display rule.
So that is the outline of what the system would look like -- more than one consolidator, markets free to set the terms and conditions, of course in conformity with the prevailing regulatory structure in the dissemination of data.
As we evaluated the benefits, the risks, the costs associated with moving in that direction we broke down our discussion into two distinct parts.
First, and what I would like our discussion at this meeting to focus on first, is the technology. What does the system look like in terms of the hardware, the software, the process by which data is disseminated? Are there benefits to moving to competing consolidators in terms of that? Are there costs and risks?
We are going to have two presentations that address that, which we will turn to in a second, but that is the technology issue.
Will the system work, will it break down, or will it indeed be doable with relatively manageable sets of risks? To anticipate a little bit, and I don't want to jump ahead or abuse my position as subcommittee chair, my suggestion is what we are going to hear is that there are things you could characterize as risks but depending on your perspective about the economics, the policy issue, you are going to have a very different view of those risks.
In other words, if you think the economics point to the viability, the desirability of the move to competing consolidators, my sense is you will see these risks as manageable without significant regulatory intervention. On the other hand, if you don't think the economics suggest that we ought move in this direction the risks are going to stare you in the face, and you are going to ask yourself, "Why bother to take them?"
That is why I suggested that we have a discussion of the technology and these kinds of risks, but we don't belabor it much past 10:30. There are important issues to talk about. However, I think the economics, the underlying policy question, is the real driver here, and I do want to make sure we have enough time to get to that.
With that in mind, I would like to ask Tom Demchak to give us an overview of what the technology looks like and what risks we might face in moving to a competing consolidator model, and I hope he can do that in about 10 or 15 minutes.
Mr. Demchak: I will try to do this within five or 10 minutes and then I will open it up for questions.
At the first subcommittee meeting, I conveyed to the participants of that subcommittee how the current CT/CQ system works and then four risks that we had identified in a competing consolidator model that I think the committee has to consider.
So what I would like to do this morning is just very briefly set the tone of how CT/CQ works and then talk about the risks, because I think you need to have that understanding of the systems.
The way CT/CQ works today is that we have got nine market centers that send the data into discrete systems located in Lower Manhattan and Lower Brooklyn supporting the Consolidated Tape System, the Consolidated Quote System. These are full tolerant platforms. The way we get the data from the market centers is we use TCP/IP protocol.
The market centers, when they are reporting their trade and quote information, package the information in what we call "packets." They send that information into the respective systems. The systems then validate the information to make sure that it's following the proper formats, that, for example, in the Consolidated Tape System we have a spread-checking capability to make sure that it is within a certain parameter from that perspective.
So once the system accepts the messages from a particular participant, we then validate it to make sure it is a valid symbol within our database. Once it is a valid symbol in the database, then we subsequently consolidate that particular message with all of the messages coming from the market centers. We do some value added calculations for our internal system in regards to what we store within our database, and we use that information for value added dissemination of information later on in the day, not with a particular trade per se.
Then we disseminate that information, the trade, to approximately about 66 direct connects that receive the data and we use a capability called IP Multicasting, so that we simultaneously distribute the data out of both of our facilities to our direct data recipients.
Now what is important about that is what we have identified as potential risks.
The first risk that we have identified is risk of sequencing. You could potentially have information get out of sequence in regards to the way we are receiving the data from each of the market centers.
Now there are three or four factors associated with out-of-sequence information. In a competing consolidator model you could have and you will have competing consolidators utilizing different routers -- in other words, communication infrastructure, different hardware platforms, and it will also be utilizing different internal software design.
Now what is important from that perspective is that in the hardware platform standpoint, when a market center sends traffic into competing consolidators -- for example, we used a Tandem environment, another competing consolidator might use a Unix system -- each of those systems service input lines differently in regards to time-slicing of information, the way they pull it off what we call the TCP/IP stack.
So you could have a situation where one market center is sending the same exact information to multiple consolidators and because of the way the operating environment is time-slicing the information and picking it off the TCP/IP stacks you could have information getting out of sequence for one consolidator versus another consolidator. Now we feel that that should be of minimal impact in regard to competing consolidators. The reason is because the way we do sequencing within the system is really on a symbol basis. That is what is important.
Once we get a message for a symbol in CTS we make sure that we guarantee the sequencing of that information going out in relationship to other messages coming in for that symbol.
Now the likelihood that in a multiple competing consolidator arena that you actually have several exchanges sending at the exact same time and the same packets the same symbols is probably very minimal, so we feel that in the processing of the time-slicing information, putting information off the stack, there should be minimal impact.
Another consideration in regard to getting information out of sequence is message-gapping. The communication infrastructure between the market centers and SIAC is a high bandwidth network and also from SIAC sending it out to the vendors is a high bandwidth network. You could have situations where either receiving a message from a market center or disseminating messages out to the vendors, the data recipients, there are gaps experienced on the lines. In other words, we either drop a message receiving it from a market center, or in sending it out to a vendor, a vendor might have a problem receiving the data.
In today's environment, utilizing the state-of-theart protocol that most competing consolidators would use, which is primarily TCP/IP protocol, the general feeling is the message gapping that might transpire is very minimal. In fact, back in 1997 we converted out outbound network from a bisynchronous point-to-point network to a router-based IP Multicasting network, and we noticed that there was a decrease in the number of message gaps that the data recipients were actually experiencing, so it is actually in today's environment a more robust environment than in previous point-to-point environments.
Mr. Demchak: -- the third consideration could be attributable to the way a competing consolidator has actually designed their own internal software within their systems.
I had mentioned that in CTS we guarantee the sequencing of information once we get it, acknowledge it back to the market center we received it and then process it through the system.
The way CTS is designed, we actually have 73 different paths within the software structure of CTS. And that's based upon on simple distribution.
The reason we have so many different paths is because we noticed several years ago that we were getting skewing in a particular symbol.
In other words the over all system utilization would be fairly low but because of a break-out in a particular stock or industry we would all of a sudden get a high concentration of trades coming through for a particular stock like Motorola.
So in the CTS world we have the capacity of about a thousand messages a second. And we actually have designed the system to handle a hundred messages per second on a symbol basis, to handle the skewing situation. So we've set up these 73 different paths.
Now, what could happen in competing consolidator, each competing consolidator would set up their own infrastructure potentially differently so you could have out of sequencing occurring based upon the way the internal system design occurs.
So the first major risk that I've just talked about has to do with out of sequence information.
The second risk that we have identified is what we call validation tolerances. When I was describing the way CTS works, we receive the messages from a market center. We validate them. And the competing consolidator mode or model, you could have multiple consolidators validating information differently.
So we see, for example, competing consolidator might reject a message based upon the validation criteria they're using.
In addition there are a lot of calculations done -at least in the CTS system and in the consolidator quote system for the national best bid and offer, for example.
We have criteria we use for calculating that information. You could have in the competing consolidator world where one consolidator is using a different calculation for the NBBO versus another consolidator.
We use price, time and size as the priority scheme for calculating an NBBO. So that's the second consideration is dealing with the validation tolerances.
And one way to mitigate that particular concern is to set up some type of standards committee, or to set up some type of organization that would -- assuming there is a continuing requirement to have that NBBO, that would set up a standard of what that calculation would entail and other calculations in the industry.
The third major consideration in dealing with a competitive consolidator model has to do with capacity. I'm sure you're all aware of the need to have adequate capacity to accommodate the traffic growth that we've all be experiencing over the last five, 10 years.
There are many, many different factors that we look at when we're dealing with capacity implications. First off, you got the network. You have to make sure that the network we have in place can accommodate the traffic requirements. Then you also have to make sure that the communication infrastructure coming from the market centers to the consolidator is adequate enough to accommodate each market center's capacity requirements.
The third component has to deal with internal system capabilities. I mentioned in the consolidator tape system we actually have threads that allow for us to support up 100 messages per second per symbol.
You have to make sure that there's adequate capacity to accommodate not just the overall industry's requirements but a particular breakout situation that might occur.
Another consideration is how much storage you have in your system, memory, the size of your database to make sure that you can handle the number of transactions and the number of symbols that you might have to be dealing with.
All these are factors that go into the equation for determining whether there's adequate capacity in a system. If any of those factors do not have adequate capacity and the demand is greater than the supply then we're going to potentially have queuing within the system.
One way again to look at in a competitive consolidator model to make sure that there is adequate capacity is to deal with some type of standards that would be put forth, or the competitive marketplace could determine whether or not a consolidator has sufficient capacity.
Because if you find that one consolidator is queuing a lot, then the industry probably would go to another consolidator to get the information.
The fourth consideration that we had looked at in competing consolidator model is kind of -- I've already talked about it to a certain extent -- and that has to deal with protocols and data formats.
Right now we have standard discreet message structures that we use in receiving data from the market centers and disseminating the information out to the data recipients.
I mentioned we're using a network-based protocol -TCP/IP in receiving the data and a distribution network based upon in IP multicasting and distributing the data.
In a competing consolidator model there should be some minimum standard set so that we have commonality to a certain degree in the way the data is distributed.
From a data format standpoint, there could be a standards committee that could be set up to deal with how we handle basic message structures -- for example, if there's going to be an NBBO required, how that should look.
And there are very effective examples that are out there in the industry such as the fixed committee which is the financial exchange committee that deals with format structures between institutions worldwide and distributing order flow information.
They're not currently being used widely for market data information, but that is a structure that has been out there for many years.
So those are the four considerations that we felt potentially could create some risks in a competing consolidator model. And as I went through each of those considerations, the general feeling that we have is that they are manageable risks based upon whatever economic model needs to be addressed in putting forth the competing consolidator scenario.
I open it up for questions now.
Mr. Seligman: Tom, for our record, can you identify the position you have with SIAC?
Mr. Demchak: Sure. I'm senior vice president in charge of one of the development organizations at SIAC. I also have the national market systems responsibility under me, which is the development side of the shop and also the product planning and operational side of the shop.
Mr. Seligman: And again just to highlight the conclusion you reached, you don't see any technological reason why a competing consolidator model isn't feasible?
Mr. Demchak: Well, I've identified the risks associated with the competing consolidator model. We feel that those risks are manageable as long as they're well recognized in the industry and that there are certain mitigation steps taken to deal with some of those risks -for example, potentially having some type of standards committee to deal with message structure changes or deal with capacity requirements associated with what we need to process on behalf of the equity and the options marketplaces.
Mr. Seligman: Appreciate that. Are there other questions from the committee? Yes, Bill.
Mr. Harts: When you were talking about the sequencing and message gapping issues, I want to make sure I understand correctly. Are you saying that the quote and trade stream that you disseminate is not necessarily on a first-in, first-out basis?
Because you were sort of implying that maybe you aggregate on a per symbol basis?
Mr. Demchak: Well, what transpires is we have connections from all nine market centers into each of our sites. We have consolidated tape system in one site, consolidated quote system in the other site. And they back each other up if we have a system or a site disaster.
The market centers, based upon the efficiency of their market systems in regards to the way they block the messages, accumulate trade and quote information from their respective trading floors.
They have a market data system that gathers that information and then they would be submitting it SIAC -separate logical line for the consolidated tape system and a different logical line for the consolidated quote system.
The way the system is structured -- I'll CTS as an example -- we use a tandem platform. The tandem platform has -- the operating system determines how it services those lines.
And it basically does it on a very quick round-robin approach.
For example, the New York Stock Exchange might have two input lines. The Cincinnati Stock Exchange one input line, the Philadelphia Stock Exchange another input line.
The system would very quickly go through those input lines and pull off the messages that have been blocked coming from the market centers and then processing them.
Once we pull them off what we call the TCP/IP stack, and we safe store them at that point, then they go through sequential processing through the system. We consolidate them and process them after we validate them to make sure it's a valid symbol and it meets the valid format criteria a market center needs to use in submitting the data to us.
Mr. Seligman: Are there other questions?
Mr. Roiter: Thanks, Joel. This should probably wait until Michael has presented his piece. But at some point I would like Tom to address what I see as dissonance between his conclusion that the risks appear manageable and at least the perception that I get from reviewing Michael's survey of vendors that the risks are not as Tom summarizes them.
Mr. Seligman: Tom, have you had a chance to take a look at the written submission that Michael prepared?
Mr. Demchak: Yes, I've read Mike's submission, the summary of the vendor. In fact, Mike had conversations with people on my staff in preparing that document.
Mr. Seligman: I think Eric characterized in crisp fashion the basic conclusion in Michael's report. How do you respond to the kind of concerns the vendor community is articulating?
Mr. Demchak: Well, they're very valid concerns. And I think I've articulated that we're also concerned in a competing consolidator model that those risks also exist.
You look at the way the industry is dealing with the path of dissemination of market data, once it leaves SIAC, you have -- as I've indicated -- 66 CTS direct recipients connecting to us to get the data.
They then provide the data to their end-users. We have no idea how they actually are providing the data to endusers and these risks that I've articulated probably exist in that distribution side of the equation.
In addition when you look at some of the risks I went through -- and I'll use the validation tolerance risk and the potential inconsistencies in the way calculations are done such as the such as the NBBO.
The CQS system gets all the information from the market centers and on a symbol basis generates the NBBO. We know that a lot of the market centers in receiving that feedback from us, actually regenerate their own NBBO.
And they do that based upon the root quotes that they're getting from us. And the reason they do that is for ITS purposes. They basically calculate the best quote.
Also the vendors today are calculating for every trade open, high, low, last, volume information. We don't really provide that information. We only provide it on correction messages.
So you already have a lot of the calculations being done by the vendors today. You have the market centers calculating best markets.
So if you look at potential inconsistencies in the way some of these calculations are done, they exist today. But I think there's minimal discrepancies because there is a standard that is put forth.
And that primarily from our standpoint would be our vendor specification where we tell the vendors and the data recipients how you calculate a last, what particular trade is eligible for calculation of the last, which are not.
For example, a sold condition -- that's an out of sequence trade submitted from a market center -- we do not include in the last calculation. So that's well articulated to the vendors.
We have a table in our spec that gives them an indication of how everything should be calculated. So if you have standards set up, I think it would minimize some of the risks from that standpoint.
Capacity -- that's another major concern that we're all dealing with -- I know in particular in the options world. You know, we've just made some recent upgrades to dramatically increase the options capacity.
And the data recipients are also going through a similar type of situation in upgrading the capacity to receive the data.
I know there have been situations where the some of the data recipients have been behind the eight ball in getting their capacity upgraded. And I think that's been then subsequently dealt with by their end-users in regards to whether they're happy with the service they're receiving.
Mr. Seligman: Are there other questions from the committee?
Mr. Atkin: Eric, I think that there's really little difference between what Tom is saying and what I've found by the vendors.
And I think Don summarized it to start this conversation. There are risks. The risks can be mitigated, but the vendors say, what's the point?
And they get a good feed from SIAC and there's no real value to them in taking multiple feeds and in mitigating those risks under the current structure of NBBO and display rule.
So I think the two-minute summary -- if you don't mind, Don -- of my report, which I think is pretty much that, that there are risks. The risks can be handled.
I think Tom did a great job in outlining what they are and how they can be handled. If it's simply a matter of continuing to do what we currently do none of the vendors thought that that was useful.
SIAC does a great job. However, the people that we talked to were all the technological feed processors -- the open question is, if you change the rules, does that have any impact on NBBO?
And I think it does. So I think the real conversation is about the economic structure -- certainly the technological risks look like they can be managed.
Mr. Seligman: Are there other questions for Tom?
Mr. McNelis: I don't have a question, but I would like to --
Mr. Seligman: Sure, Brian.
Mr. McNelis: -- a little bit on Mike said. You know representing a vendor view here, we would agree that -if the status quo is maintained in terms of the NBBO rule, there's not much point in anybody attempting to reconstitute what SIAC is doing.
But in an environment where the rules were changed, where we had other options and we had the ability to build different business models, this might well be a factor.
And certainly Reuters feels that we would be able to deal with these technological issues.
Mr. Seligman: Let me just ask everyone who speaks to identify themselves when they begin speaking. I noticed our reporter scrambling madly to keep up with, if you will, the traffic flow here.
Ms. Nazareth: And also for the telephone --
Mr. Seligman: Forgive me, I didn't realize that.
Ms. Nazareth: Yes.
Mr. Seligman: Who are the two people on the phone?
Ms. Nazareth: Harold Bradley and Matt DeSalvo.
Mr. Seligman: Is it possible for me to confirm they're here? Harold, are you here?
Mr. Bradley: I am here, thank you.
Mr. Seligman: Matt?
Ms. Nazareth: One out of two.
Mr. Seligman: All right, well, that's not bad.
Ms. Nazareth: So anyway it's helpful for Harold if we identify ourselves, as well.
Mr. Seligman: Okay. Are there other questions for Tom? Other statements? Tom, thank you. I would really be grateful if you can stay near us for the technological discussion.
Mr. Demchak: Okay, thank you.
Mr. Seligman: Mike, do you want to go beyond the summary you've just made?
Mr. Atkin: Not necessarily. I think my report is pretty much submitted for the record and if there are any questions, I would be more than happy to take them. But I think the conclusion is pretty much as I just stated.
Mr. Seligman: Are there questions for Mike? In other words, as I understand what you're saying, you agree with Tom that it's technologically feasible to operate in a multiple consolidator system.
At least a consensus of vendors you spoke to are skeptical that it economically makes sense to move to that system.
But in terms of can we technologically do it, you're basically saying you agree with Tom that it is feasible?
Mr. Atkin: Absolutely, except skeptical that it makes economic sense given our current structure. If the structure has changed, the game will change.
Mr. Bradley: Can I ask a naive question?
Mr. Seligman: Now, you have to start by identifying yourself.
Mr. Bradley: Harold Bradley. And I guess, the question I have is, as I listen to the vendors' main concern, isn't their business model basically to pay what is charged by the central vendor and then mark it up to the users? And that's the business model? So it's kind of immaterial what the cost of the information is.
Mr. Atkin: I think there might be two or three things in that. One is the cost of the data, second is the cost of processing the data. Then third is do the vendors just pass on any of those two costs to their customers. Is that right, Harold?
Mr. Bradley: That's correct, Mike.
Mr. Atkin: I'll let the vendors speak for themselves.
Mr. Bradley: I mean, that would certainly speak to their -- any interest in changing the model if cost is not an issue for them, as it is for us.
Mr. Seligman: Okay. Let's start with Brian and then I saw Simon.
Mr. McNelis: Okay. I'm Brian McNelis from Reuters. Our issue is not so much with the cost of obtaining the data, but the cost of distributing the data. If we're locked into a one-size-fits-all distribution model, we can't effectively serve our customers. Customers have different requirements. Some need the tradable data, if you will, and some do not. We would like to have the flexibility to distribute on a different set of rules.
Mr. Seligman: I think I saw Simon.
Prof. Johnson: Mike, I'd like to ask about something in the summary of your report which I did not quite understand. I thought the report otherwise was very clear and very helpful. You say two things. In the first paragraph you say, "Their biggest concern is the inability of assisting in multiple consolidators to prevent price inconsistencies and distortions."
And then you say in the next paragraph, "If the existing NBBO regime would have changed, certain vendors believe that not only would data quality consideration be vindicated," and then either one. I don't understand why removing the NBBO requirement would eliminate or reduce this data quality considerations, assuming that you're trying to produce, you know, unified non-fragmented market information.
Mr. Atkin: I'm not sure if that's a given, Simon. I think the question is if you are getting data from many places to produce the same NBBO, vendors thought there was a potential for that calculation to be different. If that criteria is eliminated, then you don't have the problem any longer.
Prof. Johnson: So let me see if I understand correctly. The problem then is producing a unified coherent across-all-consolidators information. That may be tricky.
Mr. Atkin: Consistent price across all distributors.
Prof. Johnson: So, in other words what the vendors are thinking tell me if this is a correct understanding is that if the NBBO regime were ended or it wasn't such a requirement you would necessarily get some fragmentation of the market information. And that's kind of what they're -- their working assumption, if you like, for what you think the data picture would look like. And that then has implications for, you know, their not being concerned about their quality and so on.
Mr. Atkin: Well, I mean I believe that all vendors or think that the inconsistencies in NBBO could be mitigated. I think tom laid out the fact that they could. So certainly everybody would be concerned with having consistent data disseminated. In the current structure, it really just seemed not worth the trouble. So I think the point was if the structure changes it is worth the trouble because you can do other things and I think the other things are what you combine the data with, what kind of data you provide to your customers, what kind of added value data you can disseminate with it. I think Tom laid out that there is lots of data that is useful that is not disseminated by SIAC that vendors or other consolidators could combine together and send to their customers.
Mr. Seligman: Eric?
Mr. Roiter: Thank you. I just want to make sure I'm clear on this. I think I am. If two or more multiple consolidators decided, "You know what? Even though I'm not required to provide a consolidation of all market center data, I am choosing to do so and I am choosing to provide the same consolidated data as would be required if the NBBO rule was retained.
Am I correct to understand that the vendors in that situation would say that the risks associated, the technological risks associated with that are manageable?
Mr. Atkin: Let me make sure I understand your question. Multiple consolidators try and deliver a consistent NBBO quote. Right?
Mr. Roiter: Yes.
Mr. Atkin: There is concern that that is difficult to do because of network issues, capacity issues, sequencing and routing. I think SIAC's assurance that that stuff could be mitigated is pretty good. The vendors also thought it could be mitigated; but it carries with it a cost of managing that and they didn't see the value in doing that.
Mr. Roiter: I'm sorry. I hate to belabor this, but for those who might feel that the preferable approach would be to allow for multiple consolidators, but to retain the vendor display rule, this is a crucial question. And that is is it technologically feasible and are the technological risks manageable to go to a model of multiple consolidators where the vendor display rule is retained?
It seems to me that these issues of sequencing and validation are going to exist in an environment where you don't have a vendor display requirement if two or more consolidators choose to consolidate. And it need not necessarily be consolidate all markets, but at least a group of markets. You're still going to have sequencing questions. You're still going to have validation questions. And I for one see an inconsistency in the responses that vendors have given depending on whether the vendors display rule is retained or not.
Mr. Atkin: Well, Eric, I think the answer to that is, yes, those risks -- the answer is there are risks. Whether those risks can be mitigated or not is an open question. I think SIAC says their evaluation is for areas of risks. All can be mitigated, but require standards, coordination, and work.
So I think that's the point. That, yes -- yes, there are risks. The general assumption is those risks can be mitigated. But what is the value in doing that if you have a, you know, a continuation of the existing regime?
Mr. Seligman: I think I see Ed.
Mr. Nicoll: Ed Nicoll from Datek. As I remember, there was a discussion at the committee level when I was there to the extent of how much SIAC's sequencing now actually reflects the order that the trades, you know, were executed at in the first place. I mean we have -- since we have the single source and since that single source sequence is the trades, we have sequencing by fiat more than we necessarily have sequencing that reflects the actual order of the transactions.
As I remember, in fact, is it not true that the exchanges, themselves, don't sequence the orders. They're basically reported in the order that they're received by SIAC. Is that correct?
Mr. Seligman: Let me allow Tom Demchak to join us for this part of the discussion and respond to your question.
Mr. Demchak: That's correct. That was the discussion we had, that the nine market centers accumulate the data within their own environment however they do that. When I went through my evaluation here, I mentioned about sold conditions. There is a requirement, for example, in the CTA plan that if a trade is reported, 90 seconds or more than you have to denote it as the condition in sending it to SIAC.
Now in regard to the actual point of execution or the issuance of the quote and the gather of information on the floor and issuing the quote. From a consolidator's standpoint, we don't know when that occurs. We do have a time stamp of when we received the information from the market center. And that's when their market data system basically transmitted it to us.
What happened south of their market data systems down to their point of sale and the actual trading floor environment, the activity that goes on there, we have no indication or understanding of when the actual execution or generation of the quote occurred.
Mr. Nicoll: So, if I could just -- one of the issues, the real world issues here is that, you know, professionals use the time stamp that you put on a transaction as the time when the transaction actually occurred. It was commonly thought that that is, in fact, the time when the transaction occurred. It's important.
Mr. Demchak: The time stamp that we generate on our trade line and quote line is the time stamp at point of dissemination. So we have already received the information from the nine market centers. We have gone through the validation exercises. We have gone through all of our internal processing. We've consolidated information and now we're putting it out on the outbound dissemination network. And that is at the point in time we add the time stamp. That is clearly defined in our specifications that the time stamp on all of our records is at point of dissemination not reflective of either the input time stamp received from the market center or the actual time that it occurred on the market center's floor.
Mr. Atkin: And the queuing you're talking about is the input queuing; right?
Mr. Demchak: Yes.
Mr. Nicoll: Okay. I hate to belabor this point, but there is something commonly referred to as sort of time in sales. People go back to the time in sales and they look at the time that an execution occurred. When they're looking at that time, is that not the time stamp of SIAC?
Mr. Demchak: That could be one of two time stamps. We put the CT/CQ dissemination time on our outbound records I think it was about six or seven years ago. And that was actually at the request of the ITS Operating Committee. Prior to that, we did not have a time stamp. We worked with the vendors, once we introduced that time stamp to get them to start utilizing our time stamp. We don't really know which time stamp they're utilizing. It could either be our time stamp or their own internal time stamp of when they received the message, when maybe they put it out to their end users -- you know, I'm sure that each day the recipient is using a different time stamp reflective of what they feel is appropriate to provide to their end users.
Mr. Nicoll: Okay. I just think that this is an important point. So, in fact, what we commonly refer and what investors commonly see as time in sales is either SIAC's time stamps or another layer in between them and the time the transaction actually occurred, which would be the vendor's time that they received. That or the vendor's time that they actually sequenced the transaction and put a time next to the transaction.
Mr. Demchak: That is correct.
Mr. Nicoll: Is that what you're saying?
Mr. Demchak: That is correct.
Mr. Nicoll: And so that up until today for a long period of time -- and it is generally well known by professionals that there is a certain difference between when the trade actually occurred and when these times -- and the times that are on these time stamps, so that we tolerate a certain level of uncertainty on a daily basis even with a single consolidator at this point.
Mr. Demchak: Correct.
Mr. Nicoll: Thank you.
Mr. Seligman: I want to turn to Bill Harts, but let me just make one more follow-up to Ed's line of inquiry. What is your sense of the time intervals between input and when you disseminate data and when a vendor might post a time stamp on it? Are we talking about seconds or are we talking about minutes?
Mr. Demchak: In the consolidated tape system based upon some recent measurements that I had done prior to this meeting, our internal dwell time -- and that's from the point in time we receive the message to the time we disseminate it averages anywhere between a second and a second-and-a-half. That's our internal system dwell time.
Now, I cannot give any indication of what the requirements are from a market center standpoint of how long it takes them to actually gather the information, the execution report, submit it into their system and then the internal dwell time within their system, I don't know that. Nor do I know what the vendor's dwell time is. The only thing I can tell you is the internal SIAC system time.
Mr. Seligman: I would be grateful if you would submit a brief memo reflecting that.
I actually think Bob was before Bill.
Mr. Colby: I just had two technical questions. Bob Colby. Tom, first question is what do you do to prevent any sort of systematic biases with respect to sequencing when you get the order from different markets?
And the second is what reason would a vendor have to put a separate time stamp and to resequence the data once they receive it from you?
Mr. Demchak: Well, I'll deal with the second question first. This is really an intuitive comment on my part. There could be some concerns on the data recipient end if they use our time stamp. It could give their end users an indication of the dwell time within their own systems for processing.
So, for example, if we're putting a time stamp at 11:00 and they use our time stamp, and by the time it actually gets displayed in the end user's terminal, it could be 11:00 in 5 seconds or 8 seconds. And, in fact, one of the reasons the ITS Committee back I believe it was about five years ago asked us to put time stamps on the records is because several of the market centers were using multiple vendors on their floors and they were finding a lot of inconsistencies between the time stamps that the vendors were putting on their time in sale messages for research purposes for ITS trade-throughs. And that was the genesis of this TC/CQ time stamp.
Bob, can you repeat the first question?
Mr. Colby: How do you manage to make sure there's no bias on incoming flows?
Mr. Demchak: Oh, bias. Okay. We have for the consolidated tape system and the consolidated system dedicated input lines from the nine market centers. Those are basically high band width T1 lines coming into us or T3 lines in regards to the size of the pipe.
For each of those particular lines, the market centers assign what we call logical circuits. So, for example, you might have one physical pipe supporting the traffic from one of the exchanges and then they would assign logical circuits within that pipe for trade information, for quote information and then some of the options exchanges also use that pipe to send the options data to SIAC.
So we are receiving all of the market centers information through these pipes. They link up to our communication infrastructure network and then subsequently that network hooks up to the respective systems at SIAC, whether it's the consolidated tape system, consolidated quote system or the options systems, options price reporting authority system.
Let me use CTS as an example. We've got the nine market centers sending data into us. They block those messages in what we call 1000-block bite messages. The Tandem Operating System which just happens to be the hardware for the system that we utilize determines, based upon some time slicing of how it pulls information off of each of those input lines from the nine market centers when there is traffic being sent.
The protocol determines whether there is traffic to be serviced or not. So we just basically pull the information off on a very quick round-robin basis. And that just happens to be the way the Tandem Operating System structures it.
Once we get the information then we would -- what we define as a safe-storing routine, to make sure that we have the information, we acknowledge the information back to the market center. Once we safe-store that information, then it is sequenced throughout the rest of the system with every other market center's information, because we safe-store all the information, basically simultaneously as we pull them off the stacks. And then we process them on a symbol basis through the system based upon the respective threads that we have established in the system. And then those threads are affiliated with outbound what we call IPA logical groups going out based upon letters of the alphabet. So, we use the first letter of the alphabet as a range determination for the dissemination of data. And that's how we guarantee sequence. It's on a message-by-message basis from the market center.
Mr. Seligman: Bill?
Mr. Demchak: Did that answer your question?
Mr. Colby: Okay.
Mr. Seligman: Bill Harts?
Mr. Harts: Bill Harts. Thank you. Bill Harts, Salomon Smith Barney. Actually, as I'm sitting here listening I suspect I'm not the only one thinking that maybe a regulatory change that we should recommend is that the exchanges transmit the actual time of execution to SIAC which could then possibly pass them on to the vendors and ultimately to the users. I mean, you know, some kind of synchronization with the Naval Observatory or something like that as we do today for ticket stamps on every trading floor.
But the other thing is, as I was listening to you earlier about this sequencing problem, I think I'm sort of coming from the same place as Ed Nicoll that I was concerned that if we have a system of multiple consolidators that possibly, in your words, we could mitigate the risk of essentially having different NBBOs at the same time from different consolidators. However, based on what you're saying, I don't think there's ever going to be a way that we could eliminate that entirely. I think we're basically -- by saying this we are accepting the fact that there will be different NBBOs at the same time.
Mr. Demchak: Yes, I definitely agree with that. As I've indicated, I think that the way we do the process, it's on a symbol basis and there definitely still though is a potential risk here where you could have in the NBBO calculation, we might be pulling off a message and IBM from the Boston Stock Exchange and another message from the New York Stock Exchange as one consolidator in that sequence and another consolidator can be doing the reverse based upon the way their system happens to be servicing those lines. And, at first, I thought that that would be unacceptable.
And then I started thinking that, well, maybe we're not sequencing by fiat now, but it certainly seems that we are frequency by the rims of the designer of the Tandem Multi-tasking algorithm. And I would almost guarantee that the designer of UNIX's multi-algorithm would have a different idea as would Microsoft's and TD designer and so on. And possibly, in a competing consolidator environment, maybe one consolidator would want to place more emphasis on a particular market center. Maybe the primary should get more processing resources. Maybe they're generating more quotes or more trades or so on.
Mr. Demchak: Well, actually, the way the system is set up today is, we have both on the in-bound side and outbound side, throttles where we control the data to make sure that one market center does not flood the system.
With today's open network technology, you might have a system problem on the market centers, then, and they can flood the system.
So we've actually, at the direction of CTA, put in what we define as a dynamic throttle on the inbound side, and currently those throttles are set based upon the last capacity planning process that CTA went through several years ago which, because of DOJ considerations, is no longer being followed, but it does allow a market center to go above their level of capacity and then, if any of the market centers are experiencing, or the system is experiencing queuing, we then throttle everyone back to their approved capacity levels.
So there are mechanisms that could be put in place to manage the traffic flow in a system, and then also on the outbound side, we have throttles to make sure that we send out the traffic at the agreed-upon maximum capacity level, because again, we can flood the vendors based upon several factors.
For example, if we're having a system problem independent of what the capacity of the system is, we can be queuing a lot of information up on our end and then, once we resolve that problem, we could flood the vendors a lot higher than what we've actually indicated the capacity requirements are.
So we build throttles on both ends of the systems, CT and CQ, to kind of control some of that traffic allocation and to make sure that there is adequate capacity for everyone, based upon what they've given to us as their requirements.
Mr. Seligman: Are there other questions for Tom?
Mr. Demchak: Did that answer your question?
Mr. Harts: Yes.
Mr. Seligman: Mark?
Mr. Minister: Yeah, and I think -- Mark Minister -- I think I'll pick up where Ed and Bill were going, a little bit.
Thirty minutes ago, I would guess, if we put all the major market vendors' equipment here and picked five stocks and froze them -- I picked 30 minutes ago because the market is more active in that first 20 minutes -- we would have seen different last sales, different NBBOs, different highs, lows, VWOPS, for certain, on every one of them today, with a single consolidator.
We do this all the time. We know that. We would see different time stamps. Some people would lose sequences.
I can only envision that with multiple consolidators, it multiplies this problem, even more so, so that even with the elimination of an NBBO, you're still going to have market data discrepancies magnified by the number of increased consolidators.
So I guess I'm an acceptor of the system with its weaknesses. We use tandems, too, so we're more comfortable in our order processing.
So I guess that's more of a statement than a comment.
Mr. Seligman: Appreciate the statement. Are there other statements, questions, or comments?
Mr. Roiter: I actually thought Mark was going in the opposite direction.
Mr. Seligman: This is Eric Roiter.
Mr. Roiter: This is Eric Roiter, Fidelity. I thought you were going to say, "Well, we tolerate a little bit of discrepancy now, so if multiple consolidators just introduced an incrementally greater degree of modest discrepancy, then we could tolerate that as well."
I think the real challenge is to try to quantify the amount of discrepancy.
To me, it's comforting to know that the markets operate as effectively as they do with a modicum of deviation using a single consolidator, so it's no -- it doesn't persuade me at all to be against multiple consolidators, simply to observe that there may be some degree of greater discrepancy.
I have to know how much greater and at what level do we reach a threshold of intolerable discrepancy.
Mr. Seligman: I see Bernie.
Mr. Madoff: Yeah. Well, I'm -- the more I listen to this conversation, the more concerned I'm getting about these inconsistencies of the information.
We're going into an environment almost immediately where firms' livelihood and market centers' livelihoods are going to be determined by the statistics that they put out as to how trades are executed, from both speed as to general quality of these trades.
I'm not suggesting at all that any market center is going to knowingly, you know, do anything to improve the statistics.
However, I think the information that we're all going to be dealing with, it's going to be more important than ever that this information can be relied upon and can be accurate.
Quite frankly, I think that a lot is going to be determined just by one second, the difference between one market center's execution and another.
So from my standpoint, I think it's more important than ever that these inconsistencies be eliminated, if at all possible, and I don't know that that's possible, but certainly, I don't want to see -- I don't want to go into an environment where there's going to be more inconsistencies.
I think that some of us are hearing this for the -you know, a little bit trouble information here -- for the first time, quite frankly.
It never really was a problem when everybody was sort of dealing with the same statistics, so, you know -- but now, you're talking about a whole different environment, and I'm not sure that I understand it all, what the consequences are, but I will tell you that I'm not -- I'm leaving this conversation somewhat more uncomfortable than I came in.
Mr. Seligman: Let me address that at two separate levels, and I think I want to speak to Tom a little more.
In the midst of an answer a few minutes ago, you referred to DOJ considerations, which I presume means Antitrust Division, and you suggested that they may, in some respect, have made it more difficult to collectively address practical problems.
Can you please amplify that a little bit? What are the DOJ considerations you were referring to?
Mr. Demchak: Approximately three years ago, primarily as a result of the 1997 movement to 16ths, where there were some queuing situations in the consolidated quote feed, and that was primarily because of transitional situation in moving from a bisynchronous technology to IP multicasting technology for the data recipients.
The Consolidated Tape Association and the Options Price Reporting Authority working with the Commission moved forward with putting a capacity planning process in place. They retained Stanford Research Institute, SRI, to help facilitate that process.
That process has been followed for the last three to four years. As a result of some recent direction that we've received from the Options Price Reporting Authority and CTA, it's our understanding that they are no longer pursuing that process, and in fact, OPRA is re-looking at their whole process of how they do capacity planning.
There was a point in time, for example, when the process was first developed that the market centers were allowed to look at each other's projections for the next couple of years, as they went through the evaluation on the options side, associated with the DOJ considerations. The message came across very clearly that they could no longer share projections with each other.
So as a result, the CTA, for the last year or so, has put on hold any capacity planning process and the Options Price Reporting Authority, I believe in response to a recent SEC order, is re-looking at their whole capacity planning process and their whole structure of how they do things.
So from SIAC's standpoint, it's made the capacity planning process extremely difficult. We're still using about a year-and-a-half-old projections.
Luckily, in regard to where we've configured the system versus the peaks we've experienced, we've only experienced peaks of about 40 percent on CT and CQ of where actually capacity numbers are, and in fact, we're going to be increasing some of the capacity of those systems anyway, and OPRA has authorized us, independent of what the estimates are from a projection standpoint, to do very dramatic increases on the OPRA capacity.
In fact, we're going to be later this year doubling the current capacity of OPRA from 24 to almost 48,000 messages a second.
Mr. Seligman: Let me pursue this a bit. A re the areas where you have stopped going forward with efforts areas which would eventuate in a filing with the Commission, whether it's in the form of a plan or an amendment to some existing plan?
Mr. Demchak: I don't know if I can really speak to that as the processor. That's really more the OPRA policy and CTA exchanges pertaining to that. I mean, we're not really --
Mr. Seligman: Well, let me ask Bob.
Mr. Colby: I didn't want to address that question directly.
What I wanted to say was that the DOJ order, one of the orders that came out of the options case, said that they -- that the options markets needed to stop sharing capacity planning information directly, but it left open the ability to use an agent to collect it, and use that as capacity planning, without distributing each market's individual capacity planning estimates to all the other markets.
Mr. Seligman: All right. My assumption is, having read a fair number of antitrust cases dealing with the implied immunity for securities and options markets, that there are ways, as Director Nazareth has suggested, where you can address this.
Mr. Demchak: The only point I was making, over the last year or so, it has made -- we have not been getting projections from the exchanges pertaining to what we should be configuring the systems for, and I know that, at least in the options world, they're looking to address that.
Mr. Seligman: Rick?
Mr. Bernard: Rich Bernard, if this is an intelligent comment, and it's Bob Britz it's a dumb comment. (Laughter.)
Mr. Seligman: Let the record reflect it's Rich Bernard initially.
Mr. Bernard: If I can try to put what Tom was saying together with what Bob was saying -- and this is simplifying a lot, and competing consolidators doesn't change this problem, I should point out -- the issue in the options case was an allegation that brought through to the Department of Justice settlement that capacity decisions made by OPRA had the effect of chilling competition among the markets.
In effect, they capped the system so people couldn't, who wanted to add new classes, couldn't do it.
The problem that Tom is talking about is that if Tom or any competing consolidator can have the shared wisdom of the consortium in thinking through capacity issues, that allows a more careful capacity design program.
What the import of that order is, is that you can't get the markets talking to each other to come to seem like a sensible capacity decision, and so the burden now shifts to SIAC or anybody independent, or any competing consolidator, to talk one-on-one to each of the markets.
The net result of that is to force more capacity, probably, when you get it all done, being built into the system because you have to -- you're doing more guessing, because you don't have the synthesis of the markets talking to whether, because what was really happening is, to no one's surprise, was that the primary markets would make, you know, capacity -- provide capacity information because they had more complex models, and then basically, the regionals would tend to just kind of, you know, put in their pro rate capacity projections.
So that's the problem that's been caused.
Mr. Seligman: Thank you, Rich, and that was a comment which I think should go out under your name.
Let me return back to Tom, and I want to follow up on Bernie's last statement and on Bill Harts' last statement.
We are moving towards a report that will probably be bimodal in nature, that will both focus on what improvements we would recommend if we keep the NBBO and a unitary consolidator, and what improvements or recommendations, I should say, we would propose if we moved to a multiple consolidator system.
Most of today's discussion is going to be with respect to multiple consolidators, but I want to focus on our other prong, if you will.
I would like Tom to prepare for us a memorandum which would address how we can move to reducing some of the temporal and other discrepancies he suggested exist and Ed, I think, rather eloquently brought out exist.
Bill suggested a time stamp -- I suspect it would be at the time of input rather than dissemination -- might be a means.
I'm not going to pretend to have the technical comprehension of the system at the level that Tom does, but there may be other similar issues he would want to put on the table for us that would, I think, enrich our record, and I would be grateful if, within the next couple of weeks, it would be possible for you to forward that memorandum to us.
I now have Mike Dorsey.
Mr. Dorsey: Mike Dorsey, Knight Trade Group. I just wanted to follow up on what Bernie said. Bernie came at it from more of a business side, but these inconsistencies affect the disciplinary side, as well.
We have seen NESDR reports inconsistent with vendor-type information, where their report, of course, is the cornerstone of their disciplinary actions.
Right now, there's a kind of a controversy brewing in their best execution reports, but we know it's pervasive. It goes through all the reports that they generate.
Mr. Seligman: Let me at this point see if I can make some generalizations and see how the committee feels about them.
Number one, I would infer, from a series of comments, if we stayed within the current model, which is an NBBO model unitary consolidator, the committee would favor moving as effectively as possible to eliminate these kind of temporal discrepancies.
Does anyone disagree with that?
(No response.)
Mr. Seligman: Okay. So I take it there's consensus on that.
Number two -- and this is based both on the statements that have been made to us today and on the written submissions -- I take it that, as far as technology goes, there is a general belief that there are risks and costs associated with moving to a multiple consolidator model, but nonetheless, such a model is technologically feasible.
Does anyone have disagreement with that? That is, there would be collective action issues, there would have to be ways in which you could have effective conversation without running into DOJ-type problems, there would need to be certain standardization by one means or another.
But with those kind of understandings, which I think were very well suggested by Mike and Tom's oral statements and written statements, does everyone at the table agree that moving to a multiple consolidator system would be technologically feasible?
I see Ed Joyce here, and this is in the equity context.
Mr. Joyce: Yes. It sounds to me that it would be acceptable and do-able with that level of tolerance, I guess was the word, possibly expanding, as you go to it, that on one hand, you introduce the notion that we would have to reduce the tolerance that's in the current system, and as I've understood what I've heard this morning, is extending to a multiple consolidator, would expand that tolerance among market centers.
Mr. Seligman: All right. I see Rich.
Mr. Bernard: Just to put something in perspective, although we don't see it, except in the way Michael was talking about, the issues of time sequence are far more profound within the market centers than they ever are going to be at the consolidation level or at the vendor level.j
Secondly, in the competing consolidator environment, where I at least think we're talking about a handful of competing consolidators, the issues of managing these risks are simpler than when you move either back up the chain to the markets or back down the chain to the 1400, ultimately, vendors and brokers who are disseminating data.
So while these are real issues, it's an easier problem because the number of players that would be involved is really quite small, in my prediction.
Mr. Seligman: Rich, let me move you down the time sequence of the chain, if you will.
Bill has suggested that one approach that might deal with some of the apparent discrepancies in NBBO would be time stamp at point of input to the consolidator.
Could we have time stamp at the point of the transaction?
Mr. Bernard: Well, that's easier the more electronic the exchange is, but when you get to a dispersed market like NASDAQ, time stamping, sequence, last sale rules, sold sales are far more profound issues, because the actual trade is taking place at a trading desk.
And then, I'm talking about ex-facility execution, a NASDAQ is moving toward, but when you're talking about upstairs trades, we have no idea what the delays could be from the moment a trading decision is taken by Bernie or Michael's time, and the moment it's actually input in the trade reporting.
So as I say, the more electronic the market is, the easier it is to get market center time stamps.
Mr. Seligman: Let me ask you, Adena, do you agree with that?
Ms. Friedman: I would agree that the more electronic the trade is, the faster we get the execution in there, the faster we get the trade reporter.
And the fact that we do have a dispersed market model and you have some manual trades and some electronic trades. And what we receive is a time stamp based on the time the trade is sent into us.
And I do believe that that is the case, that it becomes more difficult where you have different models all functioning within the same market.
But we are becoming more electronic, just like every other market. So that issue is lessening over time.
Mr. Seligman: All right, Bill, another comment?
Mr. Harts: No, my point was that, we'll never get perfect time information for all the reasons that people have said here. I think that the problem tends to actually be exacerbated on floor-based exchanges rather than upstairs because in upstairs trades at least there is an extremely accurate time stamp placed on that trade.
The problem here is that that information is never making its way out to the end-user. It's sitting in Nasdaq's computer or the New York Stock Exchange's computer, or wherever.
Mr. Seligman: All right, we have Mark Tellini, who is representing Schwab today.
Mr. Tellini: I just wanted to reiterate that in the Nasdaq market, we synchronize our clocks to Nasdaq daily -- 97 percent of our trades are electronic.
So from that perspective it would appear that those Nasdaq trades are at least more synchronized than the floorbased environment.
Mr. Seligman: Okay. Well, I think we've addressed the technology. We'll have some further document input. We all appreciate that this is a process where the consequence of our efforts will ultimately be reflected in a report.
And the report can reflect documents that arrive after this meeting. We'll all get potentially several efforts to comment, modify, address the report.
Let's take at this point a 15-minute break until 10 to 11:00, when we will start promptly. And then we'll focus on the economic model.
(A short recess was taken.)
Mr. Seligman: We want now to focus on the economic models that were considered by the sub-committee with respect to multiple consolidators. Let me turn to Don Langevoort to set this issue up.
Prof. Langevoort: Yeah, as I said before the economics are probably the driver in the ultimate policy question here. And our discussion within sub-committee broke down into two separate parts -- what do the economics look like if you move to a competing consolidator model but retain the display rule?
And secondly, what do the economics look like if you move to a competing consolidator model and also rescind or substantially modify the display rule?
The benefits that come from moving to a competing consolidator model are fairly standard and obvious. Whereas today we have one for a small set of monopoly consolidators, you would have competing consolidators who would have a marketplace incentive to deliver a state of the art product and succeed in the marketplace based on the quality of their consolidation work.
Therefore, you would hope that innovation would occur more rapidly and product development -- quality of product development would follow.
In addition, there would be benefits that would come from dismantling the consortia. And again, I have to be careful what dismantling the consortia means.
But remember my initial description of the model, it's one where the market centers at least have the freedom to negotiate their own contracts for data dissemination and set their prices. They will be in competition in the market dissemination business. Rewards will go to the markets that succeed. Cross-subsidization, which is possibly a part of what we face today as a result of the consortia, would disappear.
So those are the benefits even within the framework of keeping the display rule.
Should you abandon the display rule, then product innovation, vigorous competition will occur much more aggressively because the vendors can design and market their products without having to conform to a mandatory display.
Market demand will determine what is delivered to investors as opposed to regulatory initiative.
Down sides. Obviously if the display rule is maintained and you have multiple consolidators and markets setting their own prices, you're essentially saying to the consolidator you have to buy each of the markets' data.
And there is a pricing power that comes with that. We had a long discussion in the committee -- not so much as to whether there is that pricing power, but whether it's likely to be abused.
The market centers are constrained by their governance structure, representation on their boards, they are constrained by the fact that they compete for listing standards.
And listed companies don't want high prices for data if the effect of that is to limit the amount of information, or widespread availability of information.
The existing regulatory structure -- probably the anti-trust laws -- provide some dampening influence.
And so even though there is a question about the extent -- or a question about what pricing pressures would look like in this environment, that's an issue that we have to face --the question of pricing pressure within a display rule environment.
Eliminate the display rule and two things happen. First of all, any pricing power that the secondary markets may have had simply because the display rule said to consolidators, "you have to buy the data," disappears.
On the other hand, there was no sense in which the mere elimination of the display rule would affect any pricing power of the dominant or primary exchanges.
On the question of whether that pricing power would be abused, you simply move back to the discussion that I outlined a minute ago.
The other cost that deserves discussion in an environment is which the display rule is eliminated is the effect on best execution.
To what extent will investors -- whether sophisticated or average investors -- be able to execute trades with a high degree of confidence in best execution if there is not a bench mark like NBBO if there is competition for various measures of execution quality, as opposed to a single mandatory one?
Obviously that's something you can look at in two different ways. You can say we need that benchmark and regulation has to set it.
You can also say that competition for different benchmark will actually deliver a better best execution environment to the investing public.
We had a very vigorous debate on all of these issues and no consensus on their resolution. And as I said before, it's probably this set of issues that really should be the driver on the full set of questions.
Now, a predicate to a lot of what I just said is an understanding of the display rule and the best execution obligation.
And I guess, we would like to turn to Annette who might provide a little bit of background as to at least what these obligations and rules are.
Ms. Nazareth: Right. I would be happy to do that. Obviously we thought it would be helpful before we engage in a full-fledged discussion of these important issues to make sure that we all understood the landscape to be exactly the same.
So I thought I would discuss the display rule and the best execution obligations of broker-dealers. I'll begin with a summary of the display rule and the rationale for its adoption.
Under the display rule a vendor or broker-dealer that provides quote and trade information to customers has to include in its display two specific pieces of data -- a consolidated quotation display and a consolidated last sale display.
Specifically if a vendor or a broker-dealer provides quote information for any stock that is traded on an exchange or on Nasdaq, it has to provide either the NBBO for the stock -- in other words the best bid and offer with size from any reporting market centers and an identifier for that market center -- or it has to provide a quote montage for the stock from all the reporting market centers.
And if a vendor or broker-dealer provides transaction reports or last sale data for any exchange or the Nasdaq NMS stock, it must also provide the price and the volume of the most recent transaction in that stock from any reporting market center, as well as an identification of that market center.
And prior to the adoption of the display rule in 1980, vendors had been displaying trade and quote information in a manner that highlighted primary market data -- both in terms of format and ease of access.
And the commission had encouraged vendors to make consolidated data at least as accessible as primary market data. But for competitive reasons at that time the vendors were reluctant to do so voluntarily.
And instead, they preferred that the Commission established a mandatory, industry-wide minimum display criteria at that time.
So the Commission adopted the display rule, which in effect requires that consolidated data be displayed in a manner that is at least as readily accessible as the market data for a particular market.
And in doing so, the Commission recognized that the manner that vendors and broker-dealers distribute data impacts both market competition and transparency.
By requiring the distribution of consolidated data the Commission sought to promote market competition by assuring that markets can compete through the display of their quoted prices.
And I think Simon will talk about this a bit later. It certainly enables newer markets to enter the market and to compete for market share on an equal basis which would probably not be the case if all of the markets' data had to be distributed.
In addition, the display rule promotes transparency by assuring that market participants are apprized of basic pricing information for securities.
Many believe that this core data, namely the prices at which other trades recently were executed and the best quoted price currently available in the national market system is essential to permit an investor to make an informed investment decision and a broker to choose the best market.
This information also assists investors in monitoring the executions that they received from their brokers. So as you consider the display rule I also wanted to note that you might recall from prior meetings that we've discussed a number of things about the display rule.
We've not only talked about whether it should be retained or abolished, but whether in it's current form, it is sufficient -- whether in a decimalized environment, the kind of information that you get under the display rule is sufficient, and whether if further information is required because of the lack of depth at any particular quoted price, whether that additional data, we should let market forces determine what that additional data would be or whether the display rule should even be amended to require additional data.
So you might want to recall those discussions as we go forward.
I thought I would also give you a brief overview of best execution obligations.
Best execution -- the duty of best execution requires that a broker-dealer seek to obtain the most favorable terms reasonably available under the circumstances for customer transactions.
And while, as I'll discuss, best execution encompasses a number of factors. Undoubtedly one very significant factor in the best execution analysis is the price of a security.
And the NBBO has served as a critical tool for broker-dealers in satisfying their best execution obligations.
As you probably know the duty of best execution derives from common law agency principles and fiduciary obligations and it forms one of the corner stones or market integrity.
And as an agent, or fiduciary, a broker-dealer has an affirmative duty to obtain the most advantageous terms for a customer's order.
Best execution concepts have been incorporated over the years into enforcement actions under the anti-fraud provisions of the federal securities laws, as well into various SRO rules.
And in encouraging the establishment of the national market system, Congress and the Commission gave a prominent role to best execution concepts.
In fact, one could argue that each of the five statutory objectives of the national market system -economically efficient executions, fair competition, price transparency, executions in the best market and the opportunity for executions without the participation of a dealer, each of those emphasizes the importance of best execution in maintaining fair orderly markets and protecting investors.
As you know, best execution is a facts and circumstances analysis. And it encompasses a number of factors, starting with the price of the execution and the opportunity for price improvement. Also relevant are execution speed and with limit orders the likelihood of execution.
Other factors include the size of the order, the trading characteristics of the security involved, the availability of accurate information and the cost and difficulty of achieving an execution in a particular market.
For institutional investors certainly anonymity and liquidity might be overriding concerns, as well. In any case the quality of execution must always be viewed, as you know, from the customer's perspective and not from the brokerdealer's perspective.
And to reiterate the NBBO provides important information to broker-dealers seeking to satisfy their best execution obligations.
But as I've just mentioned and as the Third Circuit recognized in the recent Newton Case, there are several other relevant factors besides price.
The relevance of the NBBO alone may not be sufficient. And routing orders to a market that merely guarantees an execution at the best published quote does not necessarily satisfy the duty of best execution.
Broker-dealers have to regularly and rigorously examine their overall execution quality that's likely to be obtained from different markets trading in a security.
And as Bernie rightly noted earlier on, the Commission's recently adopted execution quality disclosure rules will assist broker-dealers in this examination.
But keep in mind that they're only as effective as the inputs. And one significant input under those rules is the NBBO which is used as a bench mark for determining market quality -- execution quality.
Mr. Seligman: Thanks, Annette. Appreciate that. Let me also turn to Simon Johnson who prepared as memorandum looking at some of the potential economic issues if you move to a multiple consolidator environment.
Mr. Johnson: Thank you. There is a detailed memo which is available outside and I'm not going to repeat all the details.
We can perhaps go to some specific definitions and try to clarify issues that come up repeatedly in other discussions such as precisely what we mean by market power. We can come back to that.
What I would like to do is just take a couple of minutes and emphasize a few key points in a slightly different way that I hope is going to be helpful.
I think that there are some very important economic assumptions in our discussion and I think these assumptions are somewhat hidden or a little bit implicit, perhaps, in some of the alternative models that are being put forward.
I would like to make those assumptions clear and question them a little bit.
So here's the reasoning, as I understand it, sort of the hypothetical thought experiment that's going on when somebody proposes so-called competing consolidators model.
I think there are two very, very important assumptions. I want to sort them out.
The first takes us from the step of saying, right, we have now competing consolidators, and this leads us to more competition with or without the display rule, and we'll come back. Obviously, that's a critical distinction. We'll discuss that at great length, I'm sure.
Now, that section is pretty obvious. We're calling it Competing Consolidators, after all, so competition is implicit in that, perhaps, and somehow, perhaps clearly, perhaps not to clearly, we're going to get more competition in the profusion of market data.
So that's assumption number one, which I'm going to question in a moment.
Assumption number two is that, by having more competition in the market data, let's say that happens, this is going to lead on to some good outcomes, hypothetically.
For example, innovation has been mentioned quite a bit. More satisfied customers is also, I think, more or less implicit in what people are saying. So this is with regard to quality and lower prices, perhaps.
Now, I think at this level, you know, the argument has seemed to many people in the room rather appealing and perhaps very obvious, and perhaps that's why they haven't expanded on every one of these steps, and I hope that we can have a discussion about this.
But this is just a -- this is one hypothetical construction and one set of assumptions, if you like, economic assumptions, that gives you this logical structure and this supposed set of outcomes.
I have three issues that I really want to raise, three, I think we can call them risks, each one of which by itself is a fairly serious risk that we need to discuss. I think, taken together, they are extremely serious.
The risks are like this. Each one of them is expanded on in more detail in my memo.
The first is the assumptions, the reasoning that I gave you a moment ago really assumes a regulatory structure which ensure fair and reasonable behavior on everybody involved.
I think the first risk, that's a very important risk, is that this regulatory structure and the way that regulation works, if you really look at the details, this is going to be undermined by the movement to multiple consolidators.
There's two issues, one of which I feel we still haven't discussed satisfactorily, which is about SRO funding. That's clearly very important in the existing system, and just waving your hands at that, I think is not satisfactory. The second is, picking up on what Anette just said, that clearly, in the way the regulation currently functions, the display rule, and perhaps more particularly the NBBO, plays a very important role. It's not the only thing you have to look at and it's not the be all and end all, but it's really a crucial part of the regulatory mechanism.
If you take that away -- so this is obviously in the competing consolidators without the display rule -- if you take that away, how is regulation going to function?
So that's one problem, or one risk, potential risk.
The second potential risk, perhaps a more serious risk -- we can discuss that -- is that some form of market power will develop.
Now, we've had a lot of -- there was a lot of perhaps confused discussion last time about what exactly is market power, and I've tried to spell it out in considerable detail in my memo, and we can talk much more about it, if you want.
I think that we're talking about situations in which there is price setting ability, not necessarily the ability to charge infinite prices, that's certainly not what market power usually implies, the ability to set prices to some extent, and the extent to which you can set prices obviously has various constraints.
Now, the really important aspects of this, I think, that I would want to emphasize relate to the fragmentation of information.
I think that given the discussion, the technological discussion earlier today, and given what we know tends to happen when you have relatively unconstrained market competition, I think you would have to expect that market information would fragment into different pieces.
You would not have the same sort of consolidation, you would not have -- there's a further additional assumption, which is put forward in some proposals or assertions, saying that these multiple consolidators would necessarily product the same sort of consolidation of information as exists today.
I think that that's quite uncompelling, based on what we know about how markets of this kind operate.
I think we could also have -- I think that the nature of market power would change. I think who has the market power would change.
I think you could well have the emergence of socalled monopsony power, rather than monopoly power. This is, monopsony power is the power, market power on the part of buyers rather than on the part of sellers. That would change things in an interesting way.
But I think market power as these proposals are currently construed, particularly without the display rule, I think that really pushes us much more towards fragmentation. I also think that, in terms of the consequences, remember, innovation is one of the assumptions that people are making, you get more innovation with competition.
That's not clear to me at all if you think that there's fragmentation of information, as mentioned a moment ago, if that fragmentation makes it harder for new market centers to enter, because nobody has to carry the data, nobody has to report that data. Then, you're going to actually have less entry. We tend to think that a lot of innovation is associated with either actual entry or potential entry.
So I think the step from multiple consolidators to competition is questionable, and that step, even if you have some competition of a particular kind with market power, competition not inconsistent with market power in some places and some times, whether that will give you more entry, I think is a big question.
So that's the second risk, risk of market power. The third risk, I think, is fairly obvious, and I'm surprised it hasn't come out more clearly before, is that if you're in a situation where you have this fragmentation of information, the incentive for some players, I'm sure none of whom are represented in this room, but some players to emerge who have an interest or decide that it's in their interest to pursue more short-term incentives, to put out lower-quality information or misleading information, or maybe even just to engage in actions that create the perception of problems on the part of small investors, so that small investors feel that they're not being treated fairly.
I think that this danger is really considerable, and shouldn't be ignored by us.
So there's another hypothetical. I discussed one hypothetical outcome, which is more competition, more innovation, more satisfied customers.
There's another hypothetical outcome, if I'm right about these risks, then not only may that, you know, the positive effects be less, you can actually have negative effects.
You can actually have more market power emerging. You could have less innovation. You may have -- you may well have some more satisfied big customers, but you may have much less satisfied small customers.
Now, I don't think that this would necessarily propel you into a total breakdown of the U.S. markets or anything like that. I think what you would get most likely is pressure for re-regulation, and I would urge you to think very seriously about what kind of re-regulation you would be likely to get in this situation.
I think you would get, most likely, a much more heavy-handed system of regulation than you have now.
So I really don't see either the short term or the long term consequences of this being good. Perhaps they're good for certain people, but I don't see them as being good for the whole system.
That leads me to my final point, which is what really is wrong with the current system?
I raised this question at a much earlier meeting. I've been sitting here waiting to -- trying to draw up a list of the major problems, because if we're considering radical -- usually, we consider radical change when there are very serious problems to be addressed.
But at least as people have described the current system, there's been a tremendous amount of innovation, including a remarkable degree of entry compared to other markets in other countries or in the U.S. in different markets, or the U.S. over time.
We have falling prices and falling costs of providing this market data. Perhaps they would have fallen faster. That's not clear, and I've never seen any real hard evidence on that. And we have remarkably high quality in terms of data being provided, with the caveats that have been mentioned earlier.
So if I'm correct that there are these risks that could cause some disruption or some damage to the reputation of markets, or just to the perception of markets in the eyes of small investors -- perhaps not in the eyes of professionals, but in the eyes of small investors -- why is there a case for taking these very big, I think relatively radical steps, if -- unless I'm missing something, unless you want to tell me, perhaps I've misunderstood there is something much more problematic about the current system than we've currently discussed.
It seems to me that the current system works well. It can be improved, and that's been discussed in earlier meetings, but the case for taking these steps, these risky steps towards a model of competing or multiple consolidators particularly removing the display rule, that justification seems to me to be very weak.
Mr. Seligman: All right. Well, we've had a series of introductory remarks. Let me just comment briefly on the very last point that Simon suggested.
There have been concerns with the current system that have been articulated in the recent past, going back to an initial petition and memorandum from Schwab with respect to fee levels, certainly incorporating all of the comment letters on the SEC's December 1999 conceptual release, and various submission to this group, and none of those are going to be forgotten. That is, as part of the final report, we will draw on those documents.
If we've emphasized need somewhat less than we might today, it's because we want to take very seriously those concerns, but we do operate in a framework which I think Simon has accurately captured, where we are dealing with a system that seems to work pretty well.
There haven't been, to my knowledge, so far, major questions raised about the quality. There have been falling costs.
So that ultimately, when we decide what we want to recommend to the Commission, we're doing it not with a system that appears to be broke, but with a system that we may argue can be improved or can be replaced by a better system.
But this is not a case of something that has systematically broken down. This is not a crisis response, if you will.
Now, having said that, Don's subcommittee looked at the question of competing consolidators in two different ways: first, competing consolidators retaining the display rule; second, competing consolidators rescinding the display rule.
I would suggest it's worthwhile for this group collectively to take a look at the models in those two different ways, as well, starting with the question, does it make sense to move to a competing consolidator model as suggested by Don's comments throughout today, his background memorandum, and the various proposals we have received in the course of our meetings retaining the display rule?
I would be interested if someone could make the argument as to why that would be preferable to the current system with various modifications we might consider.
Mike?
Mr. Atkin: I just want to get a point of clarification, if I may.
Mr. Seligman: Sure.
Mr. Atkin: Because Simon, maybe I misinterpret what you're saying, but I need clarification on what you mean by competition and innovation, because I don't hear -- I mean, I hear you saying competition is dependent upon more market -- more people entering the market, as opposed to competition on better data resulting from what we can do with the output from the market; and everything that I've heard you say assumes that competition is dependent upon more markets entering the business.
Did I miss something?
Mr. Johnson: Sorry. I was trying to parse out a little the statement about where innovation comes from. I think it comes from two places.
One is people already in the market who may feel competitive pressure or pressure from their Government structure or pressure from something else to change what they're doing, so to innovate.
I was also pointing out what seems to be a feature of the U.S. securities market and a feature of many other markets, which is that a lot of innovation comes from new entrants and from incumbents responding to new entrants.
I was that you have two mechanisms, and my concern would be, if you get a lot of -- if you don't have a display rule, I think you would have -- it would be much harder for the second mechanism to work.
Mr. Atkin: Harder to have more market entrants --
Mr. Johnson: Correct.
Mr. Atkin: -- but still possible to have vigorous competition in information, because you have more vendors processing data?
Mr. Johnson: Subject to my other concerns about market power, yes.
Mr. Atkin: And then the second question for clarification, market power I understood you to say meaning market power in setting market data fees, right, that's what you're referring to here?
Mr. Johnson: Well, I tried to go through this --
Mr. Atkin: Price-setting ability means price setting of --
Mr. Johnson: I was trying to go through this in a little bit more detail. In my memo, I think that that is part of it.
I think more generally we would be concerned about price setting in various parts of the -- not necessarily just at the consolidator level, but in terms of other people who may be working with data and packaging data, so the question of who has monopoly power at different parts of the production chain, if you like, where you take it going from raw data from transactions down to customers, right?
So I think it's -- I mean, I think it will be actually an extremely interesting, fascinating natural experiment, which people will write lots of papers about, even though I don't think it's a great idea.
But I think what could happen is the locus of market power may change; so people have expressed concern about certain entities having market power in the current system, those guys may well not have market power in the next -- in the alternative system.
Market power may move from this level, strictly the consolidator level, to the level of extremely powerful participants either upstream or downstream of that consolidator.
Mr. Atkin: All right.
Mr. Seligman: We, alas are proceeding under a sort of time-bound and meeting-bound charter, and now is the time to focus on a multiple consolidator system initially with a display rule.
Who would like to make the argument that would be preferable to the current system with or without various proposed changes?
Rich.
Mr. Bernard: Well, I don't have anything new to say, and I know people like Bernie have read all 300 pages of stuff that we've already written, but I'll try to get it done with a few things.
You will recall that in our December 1 model, we never made very glorious outcomes from getting rid of the consortia. They were fairly parochial, fairly pedestrian and they had a lot to do with how the markets inter-related with each other, not about more profound issues.
Let me remind you what they were.
We touched on one today already. If you're going to have the SROs getting together into a single system, you would think you would want to reap the benefit of them being together in that system, but because of the way the antitrust thing has turned out, you don't even get the ability to do a good job of minimizing capacity expenditures by sharing information.
There's no systemic reason for bringing these nine markets together, save for managing these risks that we've talked about, which I think we've heard are manageable if we care to manage them.
Secondly, I can't think of any reason, including the viability of markets, why you would ever put nine competitors into a consortium in which they would set prices together and agree on contract terms together.
Yes, there's an issue of a subsidy running from the primary markets to the regional markets, if you think that is what's happening, but why as a matter of national policy you would solve that problem by forcing them to get in a consortium together, we still don't understand.
Thirdly, those of you who don't sit at CTA meetings and OPRA meetings and the sort -- and I went to my first CTA meeting in the late '70s -- you know, with the change in market structure, those meetings have become very adversarial.
I've listed, and, you know, we've listed in some of our documents about those, but I'll tell you the most recent one, we have a very peculiar situation with our friends from the CBOE with .00001 percent of market share, so they're basically not a CTA participant in any active way, are getting about $600,000 of free market data fees every year, because they happen to be a member of this consortium.
And I just cannot make a case for why CBOE should get that benefit, but, you know, the International Securities Exchange, which happens not to be a CTA market participant, shouldn't get that benefit.
So, I mean -- and I could go on with lots of these kinds of disputes that show that the consortia have become dysfunctional.
If it's about constraining New York Stock Exchange market power of NASDAQ market power, the consortiums don't do that.
We have behaved before 1973, and I promise you we'll behave after 2001, in basically the same way, unless something else changes that I can't anticipate, which is, as I've written, more than you would like to hear it, we view fees denominated as market fees several ways of covering the costs of the exchange, and we get to those fees by going through a budgetary process and representatives of our users, and it happens to have come to out to be about 15 to 17 percent for the last 26 years, and it could get bigger.
I mean, there's nothing magic about it. It's what the users -- it's where the users care to put those.
But it has nothing to do with the consortium. The consortium slows us down, maybe, in terms of getting to market with new ideas, in terms of the number of steps one has to go through in order to change a market data fee or whatever, but it's not a positive force in terms of somehow constraining whatever market priority the primary markets have.
So I get back to, you know, people sitting around in 1975 and saying, you know, in the world of mainframe computers and that sort of thing, maybe you need a consortium to get the data of the non-primary markets out.
I'm not sure that was necessarily true in the mid1970s, and indeed, it was never required for quotes, and the requirement to do so in terms of last sales was rescinded by the SEC in the late '70s, so today the only thing keeping the consortium going is the fact that we're in a joint venture agreement, and to withdraw requires the SEC approval.
There may have been a technological reason. I don't -- no one ever made a case for subsidy to the regional stock exchanges. The reason why the CTA plan or the CQ plan was created, people really thought there was going to quote competition, for example, in this country.
In any event, whatever those reasons were in the mid-'70s, they're no longer true, and what you have is nine competitors coming together, and we just don't see the basis for forcing that to continue -- nothing more grant than that.
Mr. Seligman: Okay, appreciate that, Rich. Others who would like to speak in favor of a competing consolidated model retaining NBBO? Eric?
Mr. Roiter: Eric Roiter of Fidelity. If I could, Joel, could I preface this by asking Simon a question first? And then I'll get to it.
Mr. Seligman: Sure.
Mr. Roiter: I did read your paper. I didn't see the words "natural monopoly" in your paper and in your summary this morning, you didn't speak of a natural monopoly. Can I ask do you see a single consolidator as a natural monopoly?
Prof. Johnson: I'm not quite sure I understand the question. By natural monopoly, we usually mean, you know, that there is a single gold mine or a single source of oil in this particular place, so it makes sense for one company to run it. Is that what --
Mr. Roiter: No. I mean something different. I mean that as a practical matter there is no way to deliver a service or a product other than through a single provider.
Prof. Johnson: Well, I think that, you know, you can -- I don't think it's physically impossible and that was the discussion this morning. I don't think it's physically impossible to deliver the consolidation of data through more than one provider just as I don't think it's physically impossible to deliver electricity, more than one production, electricity production company, generating company. The question is what is the cost and benefits of doing it each way.
So, it's not a natural monopoly that in the sense that physically, it's just an overwhelming case for one person doing it. I think it's a question of economic costs and benefits of having one consolidator versus multiple consolidators. And that's what I was trying to address.
Mr. Roiter: Okay. Well, with that I think I would try to articulate a public policy for multiple consolidators. I think we are starting from the premise that it is not a natural monopoly to have a single consolidator. And I think we also have the shared premise that the technological challenges of multiple consolidation are manageable.
So then that leaves I think a question of whether to err on the side of allowing for a greater potential for competition when it's at least conceivable that potential competition could arise. And I think if we look back to other examples where we had single providers, whether they were, you know, a single telephone company, AT&T, even single cable operators in local areas, it was hard to make a case that competition would really lead to great benefits because often monopolists did a pretty good job of providing a product, a service.
I, for one, would say that if there is not a natural monopoly and it's technologically feasible to provide a given product or service through multiple providers that there should be a presumption in favor of that unless there are identified serious public policy interests that would be compromised by moving away from a single provider.
And Simon has alluded to some risks, but I, for one, am not yet persuaded that they are anything other than speculative at this point.
The SEC, let's remember, does retain regulatory oversight. If other problems arise from a multiple consolidator model, those could be addressed. I think the burden should be on those who are in favor of preventing the potential for competition to be more specific and to try to put some quantification to the level of risks to various public policies that would come about as a result of multiple consolidation.
I can't tell you that there will be tremendous benefits arising from multiple consolidators either in price or innovation, but I can't rule that out. I can't rule out the possibility that somebody could come up with a completely different model for delivering consolidated data, perhaps an advertising model where users who accept consolidated data might get it more cheaply or might get it for nothing if that particular provider of consolidated data were able to find another source of revenue that could make it profitable to provide market data free or at least at a price that is lower than a single consolidator would provide it.
So I'm still listening and I'm open to being persuaded, but I am not yet -- I have not yet seen a compelling case for preventing the potential for competition. And I would add, finally, that even if there are no new market entrants, if you remove the barriers to entry, there is potential competition. And the mere presence of potential competitors can have a pro-competitive impact even in environments where you have a single provider of the service.
Mr. Seligman: Eric, let me ask you a follow up, if I could. The question I posed was who would like to make the case for multiple or competing consolidators with the NBBO. I would like you to focus on the final with the NBBO or display rule part of it which I don't think you did address and explain why you both favor or seem to favor competing consolidators and the retention of the NBBO.
Mr. Roiter: Okay. Well, I think that is one public policy that is dealt with in Simon's paper and one that I think on balance probably has much to be said for it. I think it preserves the potential for competition from the smaller markets. I have a concern that in the absence of an NBBO requirement that the already difficult competitive position of the smaller market centers will be made even more difficult.
Much has been said and debated about payment for order flow. I think if you eliminate the NBBO rule you will see even greater development toward payment for order flow as smaller market centers have to pay simply to have their quotes looked at, perhaps, if not to actually have orders brought to their market.
And, accordingly, I think that the NBBO is imperfect, the NBBO rule; but on balance, it does at least have a pro-competitive impact. And that is to allow for the market centers outside the primary markets to at least put themselves in the position to get their best prices, their best quotes and last sales out to the wider markets.
Mr. Seligman: Thank you, Eric. I think I saw Adena.
Ms. Friedman: I'm not going to say anything dramatically new from what you have already said, but I think the crux of the issue really lies in whether the vendors would be able to comply with the vendor display rule without a single source of the data.
We do believe that the NBBO is an important piece of information to have out there. I think it's important from the firm's view to meet their best execution obligations, it's important as a benchmark for evaluating execution quality. So we don't disagree that the vendor display rule is -- the existence of the NBBO is important. But the question is over the last 25 years has technology improved to the point where we don't need the cumbersome nature of a plant and all of the issues that Rich just described that go along with the existence of a plan in order to make sure that those vendors can comply with the vendor display rule.
And NASDAQ believes that they -- that the data can be delivered effectively to the vendor community without the existence of a cumbersome plan. I think it's a pretty simple issue and it does circle back mainly to the technology issues.
Mr. Seligman: I see Michael.
Mr. Atkin: Isn't it possible to be able to get the benefits of consolidation from processing efficiency side, but change the other three things that Rich talked about which is contract terms and business models, revenue sharing and fee mitigation?
I mean you can still have a centralized processor sending out a consolidated data. You don't necessarily need to have everybody sitting around the table doing admin together and saying contract terms and -- are we talking about both of those things together as one subject? Or is that yet another alternative?
And, Rich, I think that's what I kind of heard from your presentation.
Mr. Bernard: We've always distinguished between the two. We've always said that if you -- if a compelling case is made for retaining the single tech system as a technology matter, that still was no argument for setting prices together and for entering into contracts together. So, yeah, that's the middle ground, but I don't want to completely equate it.
If you're going to have the systems together, that would make a whole lot more sense if you weren't putting the anointed facilities manager, in this case, SIAC, into the bind it's in now where it has got to -- it is at risk in making capacity -- in effect in making capacity decisions because it's got to enter into dialogue one-on-one with each of nine markets and then figure it all out. That's not a role they originally signed up for. I imagine they will manage it if we ask them to, but that is a non-trivial issue and I would still continue to argue that if we can manage the risk that Tom Demchak and Mike have talked about we'd still be in a better place if we could also break the consortium system.
But I agree with Mike, those can be approaches to different questions.
Ms. Nazareth: Can I ask, Rich? How would even in a competing consolidated model, how would any consolidator be able to plan, you know, from a business perspective capacity, plan unless they had some means of getting information on the kind of traffic they would be seeing through their systems?
Mr. Bernard: The point is that you wouldn't have a single point of failure, so if you had multiple people doing that, if somebody made a bad guess, you wouldn't bring down the whole national market system.
We had actually -- I was talking to Eric about this -- we never executed, but we talked to the subcommittee about finding out what the switching costs were because one of the theories behind the competing consolidator model is that you don't have to get too hung up on controlling and technical specs and that sort of thing if a dissatisfied customer of a consolidator can switch to another competing consolidator. We never actually -- I guess we never got around to asking Mike to look at that. But that's part of the premise.
Mr. Seligman: I think I say Mark Tellini.
Mr. Tellini: I'd like to speak to this for one second because although we support elimination of the display rule and I don't want my vote recorded otherwise, I think the question of whether you have a competing consolidator with an NBBO or not turns on what the model looks like. And I think that's where this discussion is heading.
If we envision a model where the competing consolidators are forced to buy data from the markets on the same terms as we see today and, in fact, all of the problems we have today with data rationing, monopoly pricing, preapproval, the administrative burdens associated with those, all of those continue to imply in fact that consolidator and competing consolidator is a mere intermediary, then we don't at all recognize the benefits that could be possible from competition. If we would address those issues and, in fact, those issues could be addressed without a competing consolidator, then we're starting to talk about some real changes here notwithstanding what Professor Johnson had to say about the problems on the table. I think those problems have been well articulated in over 600 pages of transcripts and a couple hundred pages of submissions.
Mr. Seligman: Thank you. Let me by way of getting more ideas on the floor, now put on the floor the second approach that the committee addressed which is the multiple consolidators and no NBBO. Let's have a proponent of that approach or perhaps a series of proponents speak in favor of it. Somehow, I feel I should turn to Ed, if he doesn't mind.
Mr. Nicoll: First, let me address what I think needs to be said with respect to Simon's questioning of do we have a problem here.
Not to be crude, but we have about a half-a-billion dollars in rent being extracted from the financial services industry as far as we can tell right now in terms of market data fees. And there are it seems to me two groups of participants who are troubled by those fees. One are people like Datek and Schwab which question the scope of the consumers of the data which questions whether the scope of those fees are correct.
And others are participants like, for instance, the New York Stock Exchange and the Island ECN which are concerned about how those fees are allocated amongst the producers of that data.
To take two extremes, one gets a tremendous amount; the other one gets nothing under the current system. And the notion that it is not time that we look at these from both the consumer's and the producer's point of view, I think is putting your head in the sand.
With respect to our position, you know, I couldn't agree with Eric Moore. I mean what basically he's saying is where do you, you know, what's your basic policy position and where do you place the burden?
I mean our position is that there is not enough evidence to suggest that there ought not to be competition along every level that we can have it and that that ought to be -- especially in the bastion of capital markets the rebuttable presumption of the participants. And that it is for others that want to regulate heavily that automate the case otherwise.
We can't get to that. We don't see a robust competition without addressing the issue of how the consumers of the data choose to display that data to their clients. and to the extent that they can have choices as to how they display that data, that that will encourage innovation and competition along the entire chain, value chain.
We think that essentially that there are lots of questions and lots of issues to address and lots of issues of market power that exist today within the existing framework. We don't claim that by doing away with the display rule that there aren't going to be issues of market power. We just say that it is our preferred policy to deal with those issues of market power. Those are complex and we would agree with Simon that it is how the market power is exercised rather than whether it exists.
It exists today and the question is how do you deal with it?
At the end of the day, put simply, unless the brokerage firms and the mutual funds and consumers of the data have flexibility in terms of who they can and cannot purchase data from, we don't see that there is going to be the necessary competitive forces put in place to control that market power.
Mr. Seligman: Ed, let me pose a question to you. I'm a bit of a historian and I recall the origins of what is now the display rule. One of the major concerns was the regionals literally couldn't give away that data. That, in effect, if you have the equivalent to a free market system, they would simply be left off the display.
That it is not that they are not capable of competing, it is simply that the disseminators of the data would ignore them. How in a world with multiple consolidators and no NBBO could we have any assurance that the regional status still would be communicated?
Mr. Nicoll: Well, we would have to -- with all due respect, and I do mean that and I don't mean to be flippant -- we would have to examine why there's a public policy to protect the regionals in the first place and make that explicit. If that's going to underlie a policy choice, we would have to say, "Well, we think that these regional exchanges ought to be protected in some way."
But I would argue, let me argue that actually, going b