2016 Financial Markets Conference—Policy Session 3: The New Realities of Market Structures and Liquidity

What are the challenges facing market structure liquidity? In this panel discussion from the Atlanta Fed's 2016 Financial Markets Conference, experts discuss some of those challenges as well as some of the opportunities—and potential pitfalls—afforded by new technologies.

Transcript

Paul Schultz: This session is "The New Realities of Market Structured Liquidity." We will be talking about a lot of changes that have taken place in the last 10 to 15 years, particularly in equity markets. The paper presenter is Chester Spatt, who is the Pamela and Kenneth Dunn Professor of Finance at the Pepper School of Management at Carnegie Mellon University. Professor Spatt has been the author of numerous scholarly articles in finance and economics journals and is well known as a researcher. He also has extensive regulatory experience. He was the chief economist at the Securities and Exchange Commission [SEC]. He is currently a member of the Equity Market Structure Advisory Committee of the SEC. He is a member of the Shadow Financial Regulatory Committee and the Financial Economist Roundtable .

We have two discussants. Mani Mahjouri is the CIO and chief strategist of the HFT [high-frequency trading] firm Tradeworks. He has extensive industry experience, he was the global head of FX at Sun Trading, and he was the founding partner of an asset manager Meadowvale Partners. He started his career as a part of the AQR Global Asset Allocation team, and, to me as an academic, perhaps most impressive, he was a triple major in finance, physics, and math at MIT.

Seth Merrin is the CEO and founder of Liquidnet, an institutional trading network that allows institutions to trade in size in market all over the world. They have recently launched a fixed-income trading network for corporate bonds, and that contains both dark and lit venues, and I look forward to hearing their insights on what has been changing in markets.

Chester Spatt: Thanks, Paul, and let me begin by thanking the organizers for both inviting me and for also organizing such an interesting and thought-provoking conference. So, my talk is going to focus on the new realities and changing organization of the equity markets. There's been pretty dramatic changes, but pretty well-defined set of changes over the past decade or so, and it's clear that there's going to be continuing evolution in these markets. Now there's been enough time, I think, that's elapsed that is helpful to provide some perspective on how these markets have changed—where are we now and where are we going? Of course, market structure is heavily regulated—maybe more-so than any other business—due to the nature of the liquidity externality, and we've evolved to a situation where we have approximately 60 platforms that trade U.S. equities. Another obviously important underlying aspect of trading is that so much of the decision-making is delegated, at least from the point of view of the customers, the ultimate buy-side customers, whether they be retail or institutional.

Regulation NMS [National Market System] in particular brought about a lot of change to the markets, and I'll comment briefly on a few broad aspects of the statistical evidence, but much of my focus will be somewhat more conceptual: what are some of the ways in which, at a high level, Regulation NMS changed the structure of the markets; how, more specifically, the structure of incentives with respect to buy-side orders has evolved, with respect to make or take type of pricing, and some the challenges associated with that. I'll talk briefly about the interplay between some of these issues and best execution and offer some comments about speed and fast trading. I think it's also become clear, especially with the sort of evolution at the SEC, that there's an increasing role for regulatory pilots in trying to design the future of market structure.

Almost a decade ago, we had an important pilot on uptick restrictions on short sales that, I think, confirmed probably what a lot of observers thought: that basically the uptick restrictions were largely sand in the gears of the trading process. Now there's a rollout in the relatively near future of some pilots with respect to tick size, and I think likely in the next few years, potentially a rollout of a pilot with respect to make-or-take pricing. The fixed-income markets obviously have also been evolving, and now we're certainly evolving past the post-trade opacity, which characterized markets until about 15 years ago, and transparency is gradually evolving even into securitization markets. But I'll keep my comments largely focused in the equity arena today.

First, one broad comment about the nature of competition: clearly, with respect to equity trading, competition really arises in two very different ways. On the one hand, there is competition on the level of the individual order—who's going to be on the other side of that order? Who's going to provide the best possible price? But also then, of course, competition among the businesses that would like to execute the order. So, on one hand competition with respect to the customer perspective: who is going to give him the best price today for his specific execution? But on the other hand, with respect to the businesses through which one trades, how do we instill competition among these? Of course, up until about a decade ago, we had a single dominant platform. For example, in the case of the NYSE-listed [New York Stock Exchange] stocks, there was a single dominant platform—namely, the NYSE—that had about an 80 percent market share with respect to trading.

But that's obviously changed dramatically in the aftermath of NMS. I think to some extent one can think of this tension as tension between having a central limit order book and a fragmented market, and I think the question that has always confronted regulators in the equity space is, "What kind of competition do we want to encourage?" And indeed, as we look back, what kind of competition did NMS encourage? Clearly, it brought about a much more fragmented market. So at the heart of NMS, although all the rules were interconnected, what was the most visible and controversial aspect, was the so-called order protection, or trades rule, which was absolutely fundamental to it. Now, what this rule did was it basically said that each platform, when it was quoting, would be protected with respect to the top price that it was providing but that prices that were below the top price would not be protected. So if an order was going to fill otherwise at an inferior price, first that order had to execute against the best price on platforms that were offering a better price. So in effect, what this rule did was it forced the platforms to basically look to each other and potentially at least route with respect to the very best piece that would be available elsewhere. Because this was a somewhat prescriptive rule—or arguably very prescriptive—it indeed was very controversial, and this was adopted by the SEC in 2005 and fully implemented in 2007. Now this rule had some very dramatic consequences. With respect to the business of exchanges, for example, it resulted in the decline in market share of the New York Stock Exchange. For its listed stocks, its previous market share was 80 percent, today it's about 20 percent. Clearly, with respective to the issue of fragmentation versus centralization, this had some pretty dramatic consequences. Trading costs, at least for relatively small-sized executions, declined substantially, as documented in my work with Jim Angel and Larry Harris.

Trading, of course, became highly electronic. For a long time, some of the evolution along those lines were a bit stalled, but in order to benefit from the trade-through protections provided under Reg NMS, one needed to be a fast market, and to be a manual or even partially manual market, it was just not possible to be compliant. And indeed, there was no way the specialists who were taking up to 30 seconds to execute orders, there was no way that they could be compliant under the old regime. I think, ultimately, the end of the specialist system, which had been arguably an entrenched monopoly, was one of the most fundamental aspects of the consequences of NMS.

The markets of course changed dramatically, at least the spreads came down, we obviously had many more platforms, diffusion, the emphasis on electronic trading, the rise of dark pools, the nature of executions had changed. Now, this was evolving anyway, but as obviously earlier the tick size came down first from eighths to sixteenths to pennies, and all of that resulted in much smaller and smaller executions. That was a phenomenon that continued, and also with the focus on electronic trading, execution speeds became faster, quoting became more rapid, and many of the quotes were in fact cancelled. Cancellation rates became much higher. It became easier to cancel with the nature of the electronic markets, and it became arguably much more important to cancel orders that were not going to fill.

And just to illustrate some of this very briefly, I'll use a few slides that come from some of my work with Jim Angel and Larry Harris, just to point to a few of the broad trends over the past decade or so.

We saw changes in equity volume, some of which then wound up reversing themselves. Spreads did gradually decline—particularly, I would say, in the aftermath of NMS with obviously the financial crisis leading to some temporary reversals. The number of shares that are executed in individual trades obviously have fallen dramatically, and much of that fall I would sort of attribute to the period between about 2005 to 2008, which is right when NMS was kicking in. Clearly, market order execution speeds have risen dramatically. The quote-to-trade ratios increased dramatically. The counterpart of this is "what is the fill rate?" and "what is the cancellation rate?" Basically, fill rates are just a few percent—most orders that are placed, of course, are canceled, and this is a part of the market shares of various platforms. In the case of shares that had their primary listing on the New York Stock Exchange, the NYSE share of the trading volume was about 80 percent to now, about 20 percent. Finally we've had the rise of dark pools over much of this period, and I think many observers view this as a response to some of the restrictive aspects of NMS.

So that's just some background about how the markets evolve, and clearly I think it points to much greater fragmentation of the markets. So I want to turn to the question, "How did NMS lead to fragmentation, and what are some of the ways in which this led to fragmentation?" The nature of NMS focused the broker-dealers—and, to some extent, the platforms themselves because of the linkage requirements—it focused on filling by component rather than by overall execution. So in earlier eras, what one would often do is if you had some position to execute, you would often go to a platform, especially when there was a dominant platform like the New York Stock Exchange, or even, equivalently, the upstairs block market, and you would try and get the deal done as a single transaction. Well now, instead, the regulatory structure essentially doesn't really permit that. It really pushed toward executing whatever pieces you can at the tops of the book at the various platforms. And to the extent that there are a lot of platforms, inherently, at the level of individual orders, it's going to lead to a lot of splitting up of the trades, and it's going to lead inherently to fragmentation. But there's another quite separate aspect, I think, that encourages fragmentation as well: NMS protects thequotes at the very top of the book, but it does not protect quotes further down the book. So what that means is that there is no automatic routing to platforms that might be offering better prices than transaction that another platform is trying to do, except if those prices are that the very top. So what that means is there's special protection for your top price, and not protection further down the book, even though those prices might be better as well. So what does that have the effect of doing? It basically tells platforms, "Gee, if I split up my business model, and I had one business that was quoting here, and another business that was quoting here, and those were the respective tops of the book, each of those would get the regulatory protection."

So inherent in the structure is a reward to the top of the book, which to some extent provides a regulatory incentive to promote the proliferation of platforms. And indeed, the fact that the top of the book is protected but not further down, is in a way kind of an incongruity that is sort of built into the direct regulatory structure. I am not at all suggesting that a better regulation would have been to provide trade-through protection all the way down the book—that's not what I'm suggesting—but what I am suggesting is that some of the reason that there has been a proliferation of platforms is that the regulation directly encouraged it. It provides special reward to the presence of more platforms because it's protecting these distinctive tops of the books. It's giving them special status under the regulation.

So on the one side, NMS is focusing investors to get that best execution for each little piece. And on the other hand, it's rewarding platforms for being distinct and having separate operations and putting them under separate labels. So all of this, I think, has the effect of leading to greater fragmentation. Now, one could arguably say as well, "Not only greater fragmentation but greater competition among the different platform among their diverse business models." But a very different environment than, for example, the dominant platform environment of earlier eras.

Another form of this is that neither the investor nor the broker can fully manage the overall execution. Indeed, implicitly this seemed to be at the core of one of the main complaints of Michael Lewis's book, because central to his core complaint, was that when an execution came in to nearby platform, once a little piece got filled on one platform—I think he called it BATS in his story—that when one was further away, that the other quotes then basically pulled away and backed off. Now, in a sense, this is what emerges in market micro-structure models, where when executions happen and you know that it's not the full deal but there's going to be more to follow, consistent with this broad story that I'm describing, you know that the valuation of what you want to price at should move further away. So what do traders do? Not necessarily front-running, but what they do is they adjust the positioning of their orders—they cancel orders. And indeed this is sort of part of the phenomenon of what's happened since, and I think kind of a natural enough response.

I think there's interesting connections, often kind of overlooked a bit, between NMS and best execution. I think one of the indirect motivations for NMS was that there were statistics around in an earlier era which suggest that there were lots of trade-throughs of prices, but implicitly that's almost like a best-execution issue. It's not the same as best execution, and certainly not at all legally the same as best execution, but I want to just highlight at a high level some of the connections between NMS and best execution.

Best execution is basically a legal responsibility on the broker dealer to provide best execution and in some ex ante sense for his customer. This is, by the way, responsibility of the broker-dealer, not a responsibility of the platform per se. But particularly, the NMS focus is much on the platform; it focuses on the linkages between the platforms, and requires the platforms not to trade through the pricing of the other platforms. So in a sense, with respect to some of these issues, you can think of the role of the platform and the role of the broker-dealer as somewhat distinctive and, in a way, substitutes. But I do think best execution certainly is much more germane when there are more important routing decisions, and best execution itself can be distorted by incentive payments, which I think, itself, is an important issue. And that segues quite naturally into the next part of my talk.

Incentive payments have been, I think, at the core of some of the aspects of our market structure, and maker-taker and, more recently, taker-maker models are illustrative of this, so I thought I would begin with a little description of how these models and frameworks work.

So, the maker-taker approach basically involves subsidies from the platform to the maker—that's in effect the limit order provider, so that's in effect providing rebates to the limit order provider—to basically encourage the provision of limit orders and potentially paying for those, or at least indirectly charging fees, to the taker—that would be to the market order. And you could think of this at a high level as encouraging liquidity provision. And one point I want to emphasize is that this maker-taker model predates NMS. It's not that Regulation NMS sanctioned these directly. This was already the reality. This approach was already being used pre-NMS. Now, NMS provided some structure consistent with this and actually limits it to some degree, but this predates NMS. Taker-maker actually arose in the aftermath of NMS. It involved a subsidy—it flipped around the direction. It involved a subsidy to the taker. In effect, it involved a rebate to the market order. It charged fees to the maker with a limit order. So this is an attempt, instead, to enhance the attractiveness for market orders. In fact, as I reflect on it in that way, it has a lot of similarities to an older model that was called "payment for order flow," where—going back even to a quarter of a century ago, even as far back as 1990 or maybe even even earlier—we had payments made to market orders. Then, it was only to select market orders and the like, and it was a different type of approach and obviously with wider ticks and much wider, much larger payments.

So maker-taker, as I just noted, was previously allowed, but these were capped at about three-tenths of a cent per share. The caps weren't indexed, and this has been, I think, widely noted in recent years, and, of course, commissions and spreads have come down considerably over the decade. So now there are certainly calls both to index the cap—to shrink the cap and make it kind of more consistent with the current level of spreads—and also calls for elimination of this because of some of the distortions that it can create, which I will say a bit more about.

One of the sources of distortion is that the rebates and fees potentially flow to the broker-dealer who is making the routing decision. So, on the one hand, the better execution flows to the customer, but the fees and the rebates—that comes out of the broker's pocket—and the fact that there's two pockets potentially sets up the possibility that there is an agency conflict.

Now, there are some disclosures. For example, confirmation slips typically say that a rebate might have been paid, although not specifically not what happened. There certainly are some broad disclosure about execution quality, but the disclosures themselves are perhaps not so tight. NMS also includes a restriction against trading in sub-pennies. In a sense, this is one aspect, though, which allows indirectly somewhat tighter ticks than the pennies. So what about the nature of equilibrium? Well, what the consequences are for equilibrium, at least somewhat debatable, and so I'll point here to a simple neutrality theorem. And the way to motivate this is, imagine that we have some market situation in which there are buyers and sellers, or equivalently, makers and takers—although slightly different—and think about taxing one side or the other side of the market. Does it matter which side of the market we tax? Not necessarily—it depends upon what assumptions you make, but it doesn't necessarily matter which side of the market we tax. If there are no frictions, what really matters is the net trading cost. However, neutrality can fail for many reasons related to different types of frictions: transaction costs, fixed costs, and the like. And let me just suggest here a simple analogy, at least somewhat familiar to me. I am somebody who occasionally has some sports tickets to sell because I wind up with some exogenous supply because of some partial season tickets I have that I can't use, so I go to different sports platforms to sell my tickets. Well, these competing sports platforms— this is a little bit like competing platforms in the equity world— now think about the following problem: does it matter who they charge the fees to? Does it matter that they charge the fees to the buyers or to the sellers? And the answer is no, not really. If one platform charged a fee to buyers, and another platform charged a fee to sellers, what we really care about is, of course, your net price. It basically just boils down to that.

But there's a little bit of a confounding issue in the case of our trading markets, and I think of this as analogous to eBay. Suppose we had some requirement that required that if you bought goods on eBay, goods were bought at the best price. But the price didn't include the shipping fees, and you had different business models. Some allowed higher shipping fees, some business models had low shipping fees, and you kind of see that what you care about as the buyer is the net price. You can immediately understand why there is a cap on what fees are permitted—why they have a regulatory cap—because if not, there would be a big push to have even higher fees because of regulation inducement.

In terms of the platforms themselves, a platform becomes more attractive in our equity landscape the quicker the execution you get on the other side. So, ultimately, the two sides in the context of our equity markets, are closely intertwined. But this also points to a logical issue within NMS related to the nature of these fees, and arguably this leads to serious best execution concerns.  

So pointing then to the underlying agency problem: to the extent that there are different buckets of money—there's the bucket of money that is the customer's money and the bucket of money that is the broker-dealer's money—there is at least potential for conflict of interest. The empirical evidence is illustrated by the Battalio/Corwin/Jennings paper, which suggests that there is at least some scope for this in practice. So the question that arises is, "How could we potentially try to resolve this? What are conceptual solutions to this?" The work that I did with Jim Angel and Larry Harris, I think, points to different types of solutions to this type of issue, although the solutions all have various types of practical problems. But how can you solve this type of problem?

Well, one solution would be you ban the out-of-pocket fees. You ban the side pocket. You require that, basically, all the stuff flow back through the customer. So you direct the payment, basically, to the customer. Another, that you have regulations that focus upon the obligations being in terms of net price rather than gross price. Now, of course, this changes the tick size and has all kinds of complications with respect to NMS.

Another kind of approach would be much better disclosure. Perhaps more specific disclosure at the confirm level, at the trade level. Another possibility might be higher-lever disclosure, which would be less costly, which would be really good statistics in the aggregate about, "what are the routing practices?" and "what is the related execution quality?" And these all seem to be at least conceptually—if I think of this as an agency theorist, thinking about the nature of the frictions and the like—all these, in principle, would solve the theoretical problem, but they all have practical limitations. For example, obviously banning the fees changes the tick size. Now, that may be good or bad depending upon how you look at it. Directing the payments to the customers turns out to be kind of complex because a lot of these payments aren't really known until the end of the month. But that's maybe because of other kinds of frictions in the process, because the rebates aren't determined so much on a per-transaction basis but they're determined based upon overall activity. That also makes some aspects of the disclosures complicated as well.

How serious a problem is this? I guess I would say it's a little bit of an open question, because, "does the broker benefit ultimately from this net?" Well, I think it's also important to also recognize that competition for the customer business potentially may eliminate a lot of the net benefits. That is, the benefits of the rebates may indirectly flow through to the customer because even if routing is distorted, there still may be lots of competition for the customer's business. That is the nature of competitive equilibrium: it will be consistent with the agency problem. So that may cut to your perspective a bit upon how bad is this as a problem because the customers may still derive the benefits.

One supporting fact here that I'll point to is in some respects—particularly at the retail level, at the institutional level, too— is that commissions have been remarkably low for a long, long time. And it's not just that they've been remarkably low over the past five years, even. You can go back even 20 years and find retail commissions even 20 years ago that were on the order or magnitude of a couple of cents a share, or even as low as $10 to $20 a ticket even long ago, which would be kind of consistent with the story that this was a source of revenue for brokers, sure, but the customers benefited because it affected the overall nature of the equilibrium.

One other point that I want to highlight before ending this discussion of make-or-take is the issue of equilibrium with the agency problem. As you look across platforms, of course, the platforms that offer high rebates also tend to be the ones that charge high fees. This has implications for routing. Part of the implication is that part of the high rebate implies that often the brokers have incentives to route to that platform even though those platforms provide slower execution. They provide slower execution because of the basic nature of the economics. A platform that offers a high rebate has a long line, so it provides slower execution, and on the other hand, it's not so attractive on the other side. And I think this is sort of the theoretical reason why some of the empirical results have emerged as they have. So this in turn leads to a variety of policy questions, which I have kind of largely foreshadowed and pointed to in my comments to date.

Let me turn, then, briefly to the issue of speed. There's a lot of controversy in recent years about speed in trading. People complain about the arms race, and one of my favorite examples of this is in the Michael Lewis book. Michael Lewis points to $300 million that was spent to build a little tunnel for cable through the Allegheny Mountains. So I tell my students in Pittsburgh at Carnegie Mellon, "Well, why should we care if there's a full-employment act that's having people build nearby in the Allegheny Mountains spending $300 million dollars? Maybe we should sort of think that's a great thing." But more seriously, there's clearly some rents in trading, and this is a way in which the rents are being dissipated. Ultimately, would we be better served by alternate designs? In other contexts, should we necessarily ban advertising?

We know that there are benefits to speed. We can't articulate exactly what the magnitude of these benefits are, but it's clear that price discovery provides a lot of the information—information is not exogenous; it arises from trading. And ultimately speed is one way to distinguish. And indeed is differential access itself a new thing? I would say no. Co-location is not a new thing. People 50 years ago talked about the time and place advantage of being at the New York Stock Exchange. There was the role of seats and the seat prices at the exchange as a barrier to entry. Frankly, there was a lot of nepotism at the stock exchange. I sort of take that as a signal that there probably were a lot of rents from differentiated access. At one point, even, the stock exchange long ago banned cell phones. Why did they ban cell phones? I presume because they wanted to increase the value of the trading posts around the floor where you could have phones. And this was a way, in a sense, they preserved their monopoly. All of this is about differential access. Now the timescale is completely different, but I'm not persuaded that it's crucial that decisions are now faster than human decision-making. In effect, why is it an arms race now but not earlier?

 Clearly I think costs have declined over time in terms of spreads, in terms of commissions and the like. But I'm not sure, necessarily, that the arms race is a huge problem. And that's not to conclude...first, let me point to the nature of equilibrium with costs. Ultimately, in competitive markets, you're going to have participants incurring costs, and they're going to have to recover those costs. And just like we shouldn't want to have all investors being indexed or passive, we're going to have investors who use different kinds of technology. And in equilibrium, investors are going to need to be compensated. Those with higher gross returns, in many cases, are going to incur additional costs. Those are going to be compensation for those costs, otherwise we are not going to have incentive to invest in market enhancements.

Just to conclude, I want to point to a basic aspect of adverse selection in trading, which has been highlighted a lot in recent years: a lot of focus upon very high cancellation rates. In a lot of context, cancellation rates are 97 to 99 percent. Ad I think folks have lots of concerns, often suggesting, in fact, that those are inherently manipulative, and others being more cautious about that. I think my view would be more on the cautious side. What I view high cancellation rates as reflecting is that investors want to control the situation in which their order is filled. With the current technologies, they can have better control over that, and I think these can justify surprisingly large cancellation statistics, that is, high quote-to-trade ratios. Obviously, if you put orders into the market and you wanted to execute under the conditions under which you place the order and you may want to see, "Is there hidden liquidity out there? Can I fill if I place these orders?" That, I think, is perfectly sensible, but if you don't fill, you don't want to necessarily leave those orders out there because of issues of staleness. Sometimes, in fact, what happens is even the inherent act of an execution itself—and this is sort of related to my earlier comment about Flash Boys—can lead to lots of cancellations because the execution signifies changes in the state of the world; that is, the fear that traders executing would like to trade many more shares and that the initial trade is just the start. And so then, of course, pricing backs off.

Thanks very much.

Mani Mahjouri: I thought that it would be interesting to go through some experiences that I've had personally with regards to this new market. When I was looking through Chester's paper, I actually agree with pretty much everything that he had illustrated and discussed. So I thought it would helpful and beneficial to maybe go through some specific cases and scenarios where, I think, in my own experience, that the new paradigm for trading has sort of come into the limelight. I think there are some really, really significant benefits to our new system. I think that it's not perfect and I would like to share where I think there's room for improvement.

With that being said, first, I think it makes sense to kind of look at how things used to be and so I found this really, really beautiful picture. I was actually lucky enough to be able to experience this while it existed. This is the eurodollar pit at the CME [Chicago Mercantile Exchange]. One of the first trading strategies that I created that I could actually say was my idea and I worked on from soup to nuts—from acquiring data to building a strategy to actually trading it—was eurodollars. I started out working in interest rate swaps. And I was just trying to build hedges, and we found these little inefficiencies, these little bumps and wiggles in the curve. And eventually, one day I realized it wasn't that I wasn't interpolating these right, it was just that there was an inefficiency in the marketplace. So here I am with my spreadsheet and my—I guess at the time was a C-program—and I have this process that tells me, "Go out and buy these 20 Eurodollar contracts." And so everything is good until you have to figure out how to get your contracts through that. So I kind of came at it as, "Why can't I just do this on the computer?" And the answer was, "Because that's just not the way it's done."

The way this pit is set up is there are layers from the outside to the inside, and at the very core in the inside are what I was told was called locals. These were traders—basically, proprietary trading firms—that took the other side of huge, huge trades from investment banks to hedge funds to whoever really wanted to make any kind of position in fixed income. You had an entrepreneur in Chicago taking the other side of your trade for maybe five minutes, maybe for a couple days; maybe after that he got a boat or maybe he went bankrupt. It was a very tough, very interesting, but kind of beautiful era.

This is what it looks like today. A lot hasn't changed, though. A lot has. But at the end of the day, you still have all the functions that are going on there. The core function—which is the warehousing of risk and price discovery—those haven't changed. It's just the methods by which we affect those functions have changed, and I think there are some pretty big advantages.

One of the things I really, really wanted to do was to kind of figure out what today's version of this trading pit looks like. You don't really see the concept of an order book here. The order book was kind of something you saw in your Bloomberg terminal when you were kind of trying to figure out the price, but you never really had a solid, indicative price. Today you do. So I worked with this company called QuantVR to come up with a three-dimensional implementation of what today's U.S. equity markets would look like, and it looks kind of like this. In this picture, I've just plotted five exchanges—there's actually a live demo that I'll show you but before I just wanted to explain the constructs of this picture—so these are all going to be exchanges. In the "Z" dimension you have just exchanges going off into the horizon. We have about 60 places where you can trade. Twelve of those are in what's called the National Market System. They're the beneficiaries of protected quotes that Chester was referring to earlier, and in exchange they provide you with a LIT order book which basically means that for some amount of money you are allowed to see this information.

So what is this information? These are different prices, so each one of these little tiles is an autonomous piece of information. It's a promise to buy or sell a security at a certain price for as long as the order remains in place. We represented the size of the orders by the height of the tower, so a lot of these things are like little tiles, and the reason for that is that those are about 100 shares. In today's world, that is the predominant order size. I'm going to attempt to do this live.

Audience member: Who is the architect for that city?

Mahjouri: The architect for that city is predominantly HFTs. Lots of banks and hedge funds are also in this city. It's actually very interesting...city is a great way to put it. So you can really look around and this is what the order book looks like. There's just places in the order book—these giant towers—we found generally correspond to round numbers, say $80, $81. Usually we find these big banks that end up putting their orders in there.

You can see that one thing that is really, really cool about this is that all of that action that was happening on the trading floor is now captured. It is a silent picture. You had this noisy chaos and price discovery. All of those functions are still being accomplished here.

While I'm talking, this thing actually animates. You can see if you slow things down to where you're watching a millisecond every second—so you're slowing things down by a factor of a thousand—you can see its orders are coming in. They're kind of falling from the sky. The metaphor here is that if you're at the bottom of the stack, you're first in line. The new orders have to fall from the sky. Sometimes there's a trade, but that happens about 1 percent of the time, so you might not see anything here. Those are expressed as comets, so you'll see something flying across the screen. It might be really interesting.

I think the really, really cool thing about this is that as we can study this and really learn all kinds of things about the dynamics of prices that we couldn't understand before. I'd like to describe what some of those look like now.

One thing that is really important when you're a market-maker or when you're warehousing risk is this thing called queue priority. They didn't call it that on the trading floor because there was no queue. It was open outcry. But here's what it was, and here's what it is now.

When you were on the trading floor, there were certain things that really mattered a lot. And you had trading advantages. Some of those were your spot on the trading floor. When I was in Chicago running an FX business in an HFT, I had the opportunity then to talk to a lot of former floor traders and I always thought it was really, really interesting to learn because a lot of the dynamics—a lot of really successful HFT strategies—have their roots in analogs and phenomena that had always occurred. Those traders were a great source of information. They were also an amazing source of really cool stories. Your spot on the trading floor was a really, really big deal, to the point that some people said you had to defend that with your honor. And if one day you walked in and they were standing in your spot, a fight was expected; they were challenging your authority over that. And the reason it was so important was because of the people around that spot. So if you were trading frozen orange juice concentrate—which I like just because I really like the movie Trading Places—it was important to be in the vicinity of whoever was taking orders from Tropicana. If you're trading oil, it's important to be in the proximity of the Shell brokers and to build relationships there, because the closer you are to the flow, the better opportunity you have—the better your pipeline is—for getting good risk, which means taking a position you can get out of, or getting out of a position that you got into. Your proximity to the sources of information really matter.

Physical characteristics mattered a lot. I probably wouldn't have done that great on the trading floor, mostly because I don't think I'm tall enough. If you go and talk to some of the first HFTs—the founders who came from the floor—a lot of them are very imposing human beings. Much taller, and it's not an accident. If you're taller, you are sensing more information, and you are able to express yourself much more effectively in the midst open outcry.

So now it's different. All those functions are locked in those cabinets, and inside of those cabinets are just racks of servers and switches and processors moving data back and forth. But that data is the same data that was being moved back and forth in the pit. So what are the advantages now? Well, co-location is important, so instead of there being this open outcry pit, there's an order book. The fundamental function inside that order book that the exchange provides is something called a matching engine, which just matches buyers and sellers according to a very, very specifically prescribed set of rules, of which there are about 70 different versions in the U.S.

High-performance hardware and software—again, being able to very quickly get information, assimilate it, and return with something actionable. Of course, the other thing people always talk about is high-speed telecom. That hasn't changed that much either. The examples I like to talk about there are John Reuter's developing a network of homing pigeons between Paris and London to be able to share financial news. Being able to have very quick access to whatever is going on has always been important—that hasn't particularly changed.

One thing that has changed, though—and I just don't want to overemphasize this—but I think is really important, is our ability to study these things. When we talk about what was going on in the pit, a lot of this was anecdotal. So you didn't know if doing business one way was better than doing business another way. You didn't know if it was better for investors—the locals didn't even know if that was better for them. They just found some points of equilibrium where things were kind of working and it just took its own momentum and went its own direction. It was very difficult to shape.

Nowadays, you can look at things, so just for fun and because everyone had graphs yesterday and I realized I didn't have some, I looked at what would be the value of queue priority over the last 40 days on a stock that everyone knows, which is Apple. What I did was I took three million trades that had happened on INET, which is one of the three NASDAQ matching engines, and I studied where in the queue the order was when it was submitted to the matching engine. Zero means that the order who was zero formed the price. One means that somebody formed the price and you were the guy right after him. Twenty means you came a little bit later The reason this is important is that there's always a race to be first in the queue when a new price level forms. You can actually look and see precisely how valuable it is: ten seconds after, which is blue, and one minute after which is orange, and the x-axis is just cents per share. At the first five spots, you're looking consistently at about 10 mil's profit if you unwind at mid—like 10 seconds later or a minute later.

Those spots are coveted and they are very valuable, and we could say precisely how valuable they are for a stock like Apple. It's going to be different for different stocks based on their prices and other elements, but the main thing I want to illustrate is that it's really valuable to be able to ask these questions and get precise answers.

How did it used to look like? Well, here's another stock—Microsoft. Here's what I think it would've looked like in an open outcry: in the middle you have this market maker who is responsible for maintaining this order book and balancing supply and demand for Microsoft. Around him, you have all this interest. That interest, in today's world, is very meticulously organized into the limit order book, but there's no concept of that in open outcry, and for good reason: it was very impractical before technology caught up to be able to precisely mark out everyone's exact interest. But what you have is a bunch of buyers and sellers, so if you happen to be on the floor and happen to be a great trader and happen to be engaged in what's going on, you might get a sense in this case that there are more red arrows than blue arrows, and you might be able to create some kind of order book in your head where you're able to see sell interest and buy interest, and that does lead you to a bid-offer spread, of course nonindicative. When they did write that down, it was really, really wide. That's about an order of magnitude or more greater than where spreads are today for a stock like Microsoft. If you were a good trader and were engaged in what was going on, you could make money this way because you could see things like, "OK, there's a sell imbalance, and when there's a sell imbalance, my prices are going to go down." That was a viable business.

Here's the thing: there's massive information asymmetry, so only a few people who are lucky enough to gain that information are able to do stuff with it. So that limits competition. It's primal. I don't know if that's necessarily a bad thing. When we hear stories about how people interacted with one another, it's really fascinating and interesting. They were definitely fighting to get the best price for one reason or another. But it's imprecise. It's very difficult to be able to ascertain exactly how much Microsoft I can buy or sell at this moment, and that affects all kinds of trading decisions down the pipeline. It's subjective, as well. Two guys tied for a trade—the concept of tied is very subjective, and relationships matter a lot. It's opaque to us—me being the hedge fund strategist who wanted to run a zero-dollar strategy. But here's the thing that really gets me: it's impossible to assimilate information from the other stocks.

You could argue that everything above that is just a factor of the system and there's a lot of positives to the system as well, but the one thing that I can't get around is that it does take your full attention to figure out what's going on with Microsoft. And that leads to a pretty inefficient way to warehouse risk because you're so focused on the microdynamics that it's difficult to be able to get information from other stocks. So how does that look in today's world?

Now, we can go back to that picture, but when we built that city out of the orders and you look to the right and look to the left, clearly there were higher buildings on the right. I don't know if that's necessarily an indication of cell imbalance but you could imagine pretty much any indicator you want to try to understand if there's an overwhelming interest in buying versus selling a stock, and you could have those indicators in real-time for many, many stocks.

You look at Microsoft and in this chart here, red means there are more sellers than buyers, blue means there are more buyers than seller, and you can see that there's a sell imbalance on Microsoft. There's a buy imbalance on Oracle, and so what can you do? Well, you can write an algo very efficiently bidding Microsoft and offering Oracle—effectively providing liquidity to the marketplace and neutralizing a very significant amount of the risk associated with that. And you can do this is seconds—milliseconds, microseconds now—but it's a very, very powerful innovation. You can look at the relationship between the prices and you can just see that something could be behaving locally stable, and the real point I'm trying to get is that you have all this information, so can build your models even in a small, local regime you can say, "OK, for the next five minutes, if I'm long a little in Oracle and something terrible happens in tech, my portfolio is balanced." So that takes this bid-offer spread and collapses it down to the minimum, which is now one penny.
I don't want to belabor how we got here because I don't think I'm going to do as good a job as Chester did, but they way I think about it is we had new regulations. First, before you could get tighter spreads, it has to be allowed to trade at a tighter spread. The intramarket price priority—the trade-through protections—were critical because if you didn't have protections, you could have a better price all day, and what was happening was the people on the floor were just ignoring it. So it was really, really important to say, "No, that's actually against the law." If someone has a legitimately better price, you have to honor that price. There are provisions that say that now everyone has access to that market data. Whether it's equal—I don't think it necessarily is—some of it's very expensive and impractical, but it is available.

I'm going to wrap up, but the other elements are technology—so you had to have electronic markets. It would be very, very impractical to try to construct a static limit order book for a single instrument and update it live in the context of open outcry. It's viable to create trading venues. You don't need everyone to be in Chicago anymore. There are all these tools that allowed HFTs to come in and add some innovation to market-making. This was under the backdrop of very favorable market conditions, so what I'm talking about is the banking crisis, which represented an unprecedented amount of volatility and demand for liquidity. When things are so volatile, it's also quite easy to model the correlations that really matter, so we had time to learn how to do that properly—several years. Underbuilt supply of liquidity...when the demand for something is really high and the supply is really, low it's a great time to be in that business and things all kind of lined up. And for that reason, I think, we went into this new paradigm much more quickly than we otherwise would've. I think it was inevitable in any case, but I think we went really, really fast and that led to other things that I think are worth talking about.

First off, electronic trading is objective—which I think is better than subjective—but that doesn't mean it's optimal in every way. I can show you that queue priority is really valuable. What's kind of crazy though is how much money goes into asserting that first place in line. I always felt that if that was the amount of money we are spending to be first in line, wouldn't it be better for everyone if we could just make a lower price. There would be a lot less message traffic, and people would be competing on the price as opposed to the delivery of the price.

I think that things happen so fast that the first movers in the space got a very, very huge advantage, and in defending that advantage it's exacerbated the arms race. Outside of technology, there are other elements in the system that are helping to move us back into a two-tiered system from the two-tiered system we came from—that is, volume tiers. It seems like it's pretty commonly accepted practice that the more volume we do, the lower your brokerage should be. What does that do for the market? Well, some of the bad things it does is it helps the large players build a moat around their kingdom because they have a lower cost per unit. That's suboptimal, and data costs are also escalating. I think that there's definitely room for improvement along these things, but ultimately it's been a pretty interesting market.

Thank you for your time.

Seth Merrin: So I'm here to inject a little bit of debate into the session, and what I want to do is I want to talk about the parts of the market structure that are not functioning in the equity world, and then I'd like to bring it into the fixed income world, where we have some opportunity not to make the same mistakes.

First, I want to talk about how we can offer market-driven solutions, to create some solutions for this marketplace, and start with a quote from an ex-commissioner. "I cannot say it loudly enough: the industry must get together and create private market solutions— solutions that actually work and are efficient—before the government imposes its worldview."

Here we are at the intersection of two of the fastest-changing industries in the world, I think. One is the financial industry and one is technology, and when you put those two together, obviously the changes come about so quickly that the regulators have had a very, very hard time trying to impose their will, and the analogy that I use is they ask you to fasten your seatbelt way after you've already gone through the windshield. And I don't think that's a very good way for the regulators to really help this market structure. The market structure is so important, and the efficiency of that market structure is so important to the capital markets, and the capital markets are so important and vital to this country in terms of allowing companies to start, to grow, to expand. We need to focus on market structure, and I'm not going to focus on the micromarket structure. I want to focus more at a much higher level—that way we can talk about some of the problems that have been occurring.

First, for all of you who don't know about Liquidnet: Liquidnet was founded by myself in 2001 and it was created to solve a problem—a problem that institutional investors have every time they go into the marketplace. That is from 1981—from the real start of the bull market—the entire institutional sector managed about two-hundred and fifty billion dollars. Today, you've got Fidelity, you've got State Street, you've got BlackRock, that manage trillions just individually. The market itself has ballooned in the last 30 years, and what's happened is that there has become this wholesale sector, and, unfortunately, there hasn't been any wholesale market that has been created to accommodate the size of their orders.

When we saw some of those supply-demand imbalances, those were in terms of hundreds of shares. When an institution goes into the marketplace, they have millions of shares to buy and sell. What we do is we service those institutions. Liquidnet was created to create this wholesale marketplace. We now have about 800 of the largest asset managers around the world who are all participating, like eBay, where they all contribute their liquidity. Everything that they want to buy and sell; it's a one centralized, anonymous pool -- a dark pool. We cover about 44 markets around the world, which is more than 70 percent of all the publicly traded companies in the world.

Our claim to fame is that our average execution size is about 42,000 shares. Our average execution size is anywhere from 100 times to 300 times that found on other dark pool or the trading venues. It's a wholesale marketplace. The unique aspect and perspective that we have is that we've also launched six months ago a wholesale market for fixed income, and I want to talk about that in a couple minutes, too.

In Liquidnet, these buyers and sellers contribute everything that they want to buy and sell, and if there is an opportunity to match, they can cross in Liquidnet without anybody else knowing what their supply and demand was, which eliminates a lot of the trading ahead of it, of those institutions. They wind up with very, very large executions.

So, the market structure—how did we get here today? Well, with the changes in technology and changes in the financial industry, obviously we've evolved very quickly. That's fine, except that we're a highly regulated industry, as we've spoken about, and that creates distortions. In the equity world, the SEC tends to take a look at the regulations as written as if it were the Constitution of the United States, as opposed to changing with the times. They're really taking a look at this tome that was written, and most of it was written before technology was even in the marketplace. So when somebody comes and asks them or says, "I want to connect to a microwave tower so I can shave some microseconds off of my trading time," they take a look at this thing and say, "Well, there's nothing here that prevents it," so they allow it to happen. And since everything has to go through the SEC for approval, they're the ones who are actually determining what the market structure looks like.

Ultimately, what I think that we've seen is that it becomes—and it could become—a house of cards, so much so that perhaps some really well-known writer is going to write a book about it, and that's going to degrade the amount of confidence that investors have in the market. If investor confidence is eroded, and people think that the market is rigged, they start pulling out. And what happens to capital markets when investors start pulling out? It obviously erodes everything from capital formation to actually the trading of the securities.

My knowledge here is that of the tax code. How many people here think that the tax code is efficient, the one that we have today? So I'm going to ask you an A/B question: an either-or. Would you rather keep on amending the tax code, or would you rather toss it out and create a whole new one? So everyone for amending the tax code raise your hand please. Everyone for tossing it out and creating a new one. Right.

This is what we have with the regulation—that we keep on putting bandages on, and when we're talking even in this room about the market microstructure—these are just a lot of band-aids, a lot of problems that we've created just because we're just adding crap on top of crap, quite frankly.

We have 57 or 70 order types today in the marketplace. Now, I've been in the marketplace for a very long time, and I understand four: you can buy, you sell, you can sell short, and you can cover. I don't know what the other 53 order types are, but I do know that every single one of them had to be approved by the SEC. And so many of them are so singular in nature that only the people that requested that order type has the brains and computer power to be able to use it. And I would say, "Is that really good for the market itself?"

And we're going to go through a lens that I think that we should actually decide, and I think it's not that difficult to decide. Is this good for the market or not good for the market? This marketplace is really filled with a whole lot of Band-Aids, and trying to put those Band-Aids on leaky roofs, and I've just got some examples that I'm really not going to go through. What we really believe in is we should fix the root of the problem, right? We should toss out that old code and create some new code that really reflects the current realities.

Even Liquidnet, when we say, "OK, we now have a wholesale marketplace, we need a wholesale market itself." A lot of the problems in this marketplace stem from the distortions that regulators have placed on this market. Can you imagine...Walmart comes around and [Sam] Walton comes and decides that he wants to create Walmart. We have all these huge Walmarts. And imagine that if Walmart wants to sell 6,000 dozen sweaters, the only place that they can go and buy it was the corner Gap. Simply not equipped to handle that volume of business. And when you talk about the average execution size on the exchanges being 200 shares, and the average order size of an institution being above 250,000 shares, that's what we're asking institutions to do every single day. They have to go to the corner Gap to buy and sell their millions of shares, and it's simply not equipped to handle it. I assume that most of the people in this room have taken an economics course, if you're not economics majors or PhDs. If you have, on average, 250,000 shares of demand against 200 shares of supply, what happens? You're screwed, right? That's the economic term.

So you're screwed every time you go out into the marketplace—every time you have to buy and sell securities. That's a big problem, because these are the people that manage money on behalf of all of us, right? Pension fund managers, mutual funds—this money is going to pass them, but it's still managed by institutions. So you can take a look and this is where the market structure has evolved to. In the good old days, we had a couple of exchanges, and our volume consisted of retail investors and institutional investors. Investors, I think, is the key word here.

More recently, and in the recent past—since Reg NMS, really—we've seen a proliferation of dark pools and high-frequency trading. These are the two new entrants into the market structure that regulators all over the world are now really studying:is it good? Is it bad? And they should be studying it, but they should have been studying it 10, 15 years ago and coming up with some solutions or some guidelines for whether these are good or bad. But in this period of time, when we've gone from a marketplace that's 100 percent investor-driver to greater than 50 percent high-frequency trading—let's talk about the differences, but first let me put the caveat that not all high-frequency trading is bad. Some of it is, and some of it is really good. I'm somewhat agnostic. But there are issues that go on with it.

If you have a marketplace that's transformed between an investor market to a marketplace where more than 50 percent of the volume does not care about the management of these firms, does not care about the correlation of stock price and underlying fundamentals, but simply trades on the inefficiencies in the market, and their average holding period is seconds—if that long, right? You have a serious change in dynamics in that marketplace. You've gone from an investor-driven market to a trader-driver market, and that, too, creates distortions. Again, not all of which are bad, but it does change seriously the dynamics in the market, so much so, that I will tell you that—and being a global company, we deal with regulators all over the world. The regulators all over the world take a look at the United States and our market structure and equities and say, "I don't want that to happen here." So they become much more proactive, in fact, than the U.S. regulators and the U.S. market structure.

What we say is that you could either have market structure by default, or you can have market structure by design. Today, clearly, we have market structure by default, because every new request that comes into the SEC, they take a look at this—"No mention of this, seems okay, we'll let it happen"—we don't think that that is the right thing for them to do or the right way for them to regulate the market.

Here's another quote from a current sitting commissioner: "It's unclear whether dark pools continue to perform"—and I will not only criticize HFT but I will criticize dark pools, too—"It's unclear whether dark pools continue to perform the function that were originally intended." Now, I think she's being politically correct, because it is absolutely clear that dark pools are not performing the function for which they were originally intended. Dark pools were created for institutional trading, not for retail trading. Retail requires the quote. It requires the LIT, exchanges, and the exchange are truly the capitalist tool for each country. The integrity—and as a utility it really has to be maintained.

Dark pools were created to provide the anonymity that institutions need. Institutions need large-size executions. They need to be able to get a better price than what just 100 shares can provide. If any one of you were going to buy 1,000 shares of a stock, would you want to buy it one share at a time? Well, institutions—if they want to buy 250,000 shares—they, too, do not want to buy it 200 shares at a time.

These are the stats of dark pools today. Forty-plus dark pools—mostly managed by the large investment banks—and this number differs from the professor's number because it also contains the dark pool volume that is executing within the exchanges, too, and the average execution size is 217 shares. And this is consistent with dark pools all over the world. Our average execution size, again, in this country is about 42,000 shares.

Let's transition this over to fixed income, which I believe is the main reason for this conference. I'll start with one more quote" "The more we study the problem, the more we are convinced that low market liquidity is the new normal for corporate bond markets." This is one of those Captain Obvious statements, right? The big problem with the market structure—that also creates an opportunity, I would say—is that the previous market structure for trading fixed annuities, at least corporate bonds, is now gone. The large banks used to supply about $250 billion worth of their capital to facilitate these trades. There was no centralization of bond trading. You would go to J.P. Morgan, you would go to Morgan Stanley, you would go to Goldman Sachs. They would position it and they would make a ton of money on trading bonds. In fact, most of their profits over the past 10 years has come from bond trading, but with the recession and the new regulatory requirements, that capital has gone away, so much so, that from $250 billion, we're now in zero capital, negative capital. So this would be the equivalent of if the stock exchanges suddenly went away, and now everyone that traded stocks said, "OK, what do I do now? Where do I find the liquidity?" It's no different from the fixed-income market structures we have today. What they used to use in the past is now gone.

Now everybody is looking and trading corporate bonds on an agency basis. Even though the big banks were making tons of money trading corporate bonds, it was still somewhat of an efficient marketplace. Today, it's anything but. But the good news is that now we know that this market structure is evolving. In fact, it's in flux and it has to create a whole new market structure.

The SEC, which famously...one of the commissioners said, "Out of everybody that works at the SEC,"—this was about a year ago—"We have one person who's focused on bonds." I would say they should probably double that.

The corporate bond—it's all manual. Before Liquidnet came around, the only way you could trade a block of stock was through a human being. There's no way that anyone would trade a block of stock electronically. Today, in 2016— in such a large market like trading U.S. and European corporate bonds—the only way you could trade is through a human being? That has to change.

We've got a big problem. There are lots of CUSIPs [Committee on Uniform Securities Identification Procedures], and most of them are not actually changed, but since 2008, the amount of money that has gone into bond funds has doubled, and the amount of liquidity has gone from 250 billion [dollars] down to zero. Obviously, this is the liquidity crisis that I believe this conference is talking about.

We started and we created a platform for trading corporate bonds about six months ago. We're brand new at this, really, but the uptake has been enormous. What we've done is we're putting together a new market structure. Our customers owe 99 percent of corporate bonds out there, so they own the liquidity. And what we're doing is the same thing as in equities—we're creating this dark pool for them and saying, "Put everything you want to buy and sell and for the first time, we're going to centralize all of corporate bond liquidity."

See, even today, if somebody wants to buy something, they give an order to buy something, they give an order Goldman Sachs, and somebody wants to sell the same bond, they give it to J.P. Morgan, it's very difficult for those two to find each other because there's no place where they can meet, essentially. That's our value proposition here. And now we have about $8 billion of liquidity every day in our pool. Our average trade size is a little over $3 million. And again, we're still very small and this is very new, but the trajectory is hockey-sticking. We're not 100 percent the solution, but we are a solution.

Here's the lens by which I think regulators should really be taking a look at these markets and saying, OK, if somebody comes to you with a new proposition, these are the three things that it should pass through. It shouldn't be so prescriptive; it should be principle-based. So, they should say, "Hey, does it create more efficiencies in the marketplace?" If there are 57 different order types, does that create market efficiency? Two: does it promote the capital formation process, which is incredibly important to every part of this country? And three: does it instill investor confidence? And if the answer to those three is yes, then perhaps we should allow this to happen, and if it doesn't, the SEC and the regulators have to start saying no.

Not everybody can play in the sandbox nicely because if an institution has a million shares to buy and there are high-frequency traders out there and, clearly, they are going to see a supply-demand imbalance, and they start trading ahead of that institution because of the market signals that that creates, this is an example where very few people make a lot of money at the expense of very many. That, I believe, is inherently unfair because those big institutions have to buy millions of shares in order for their portfolios to perform. They can't make money 100 shares at a time. That's not their business model, and that's not what we invest in them to do.

I'm going to sign off by saying that we have a great opportunity because we know there is a whole new market structure that is coming. We've read this book because we know how the equities market went from manual to electronic and we know all the pitfalls. Why don't we just start here and try to get this right, and we can go and we can actually do market structure by design.

Thank you very much.

Schultz: We have a number of interesting questions from the audience. We won't have time to take very many of them, but let's start with a couple.

Given the many costs of complexity, wouldn't we be better off with a single utility exchange markup, such as successfully operate in many Asian countries? Chester, do you have any thoughts on that?

Spatt: I'm not sure that we would be, in particular. Now, obviously, it had a different structure. It wasn't structured as a utility. Although maybe in some ways it was structured as a utility. The old New York Stock Exchange...and I actually think that we're actually far better off now than we were with the dominance of the exchange and the specialist system and all that that brought.

Merrin: I agree with that. There are good and bad things to the complexity that we have here, but if you take a look at the margins of these monopolistic exchanges in Asia and elsewhere, their margins are 90 percent. We still have the most efficient marketplace in the world, and it's because of that competition.

Mahjouri: I think having competition is really, really important. It allows exchanges to compete on different forms of innovation and ultimately leads to more efficient transactions.

Schultz: Second question: "What are your thoughts about the risk of a crash, like the flash crash or the '87 crash, with the decline of specialists and the increase in speed of execution? Do we need circuit breakers to be reworked?

Spatt: Well, I guess my broad thought about this is that we had the crash of '87 when we did have the specialists. I know, obviously, there are a lot of concerns about how the complex structures fit together. I think the issue of circuit breakers is a difficult one. Part of the difficulty in my mind is when regulators go down this sort of path of structuring circuit breakers, the issue of how you set these circuit breakers—this is sort of easier to do after the fact than before the fact. I think it's incredibly difficult, if not impossible, to understand how to set these circuit breakers right before the fact, but yet very easy to do after the fact, which leads me to a lot of skepticism.

Merrin: This is an example of a Band-Aid that I was talking about. Now I would reframe the question, or change the answer. Any well-managed corporation has a top-10 risk list. And that changes—it's dynamic, probably every six months. You should come up with what those top 10 risks are, and then you should come up with mitigation plans for what those risks are. There are always going to be risks, and I'll give you one right now: that is that amount of money that's flowing into passive funds. There never before in history has been such a concentration of money in the exact same assets. The problem is that we've got a very, very shallow market, and what happens when lots of people are selling their ETFs or their passive funds and they're all in the S&P 500 and you have a massive rush for the exits because they have to sell those 500 securities, or some or some proximity of those 500 securities. And of course that's going to be a run on the market, which is going to whip the new entrants into the market. There's going to be some exacerbation of that trading, and you're going to have price gaps going on. So why don't we think about what those risks are, because that is a new dynamic. We've never had that before, and let's mitigate all of those risks. Circuit breakers is not the be-all and the end-all for those issues.

Mahjouri: I actually think stopping the market in the midst of potential panic and giving the fundamental forces that are making the buying and selling decisions an opportunity to reevaluate can be very valuable and useful. One of the things that technology has done is interconnect our markets to such an extent that a lot of these algorithms are developed in such a way that they are designed to lay off risk in other assets, so what you see is, in a big enough impact, as prices fall in the S&P, for instance, you're going to see other correlated assets falling at a much higher rate. Taking a break to allow trading decisions to be reevaluated, I think, is productive and useful.

Schultz: I think we're about out of time, but I'd like to ask one question that seems to be getting a few votes here. This is something that, if you do research in microstructures, you hear variations on this question all of the time. Does any of this matter for a buy-and-hold investor, or is it a side show?

Spatt: I think for the typical buy-and-hold investor, it's not so important.

Merrin: I think it's extremely important, because you have the guys who are the basis for the financial markets. They are the ones that own most of the assets, and they have to perform. If they don't perform, then people can't retire. If they can't manage those assets, then...if everything flows to passive firms, if everything comes out of the market, there's not going to be anybody to buy those IPOs or those follow-ons. So the market structure absolutely matters, and what we have to do is we have to make sure that our market structure is the most efficient market structure. It's never going to be perfect. We have to ensure—and it's the regulators' and others' responsibility—that there is investor confidence and that it's maintained and instilled in our marketplace. Anything that erodes that is a problem for the marketplace. But clearly, if somebody has to buy a million shares, and there are these little firms that are taking advantage of that supply and demand, those costs, it moves the stock price of those buy-and-hold investors. That's a tax on everybody's return that invests their money in those managers. I believe that's a big problem.

Mahjouri: I think it's really important. Liquidity is one of the fundamental tenets upon which value is built, and so if there is a perception of lack of liquidity—that does damage the value. I would like to address one of the things that you said, saying that a lot of these big institutions that are so disadvantaged. They do have a fiduciary responsibility to understand the landscape, and if you're managing trillions of dollars and you don't have the resources to understand 56 order types that a small shop with five or 10 employees and a couple millions of dollars a year of costs can do, maybe that should also be reevaluated. I don't necessarily think that complexity for complexity's sake makes a lot of sense, but I do think that when you're managing other people's money it's really, really important to understand the landscape. I don't think what we have today is beyond being able to be understood, so I think that's really important to consider.

Merrin: It could be simplified, though.

Mahjouri: It could be simplified.

Schultz: We've run over time, so, gentlemen—thanks for your insights.