2016 Financial Markets Conference—Policy Session 2: As Large Banks Back Away, Who Will Supply Liquidity?

The over-the-counter markets historically relied on large banks to provide liquidity to buyers and sellers. But changes in market conditions and regulations are reducing returns to banks that serve as dealers, leading to a relative reduction in their commitment to supply liquidity. Should we be concerned? Who will be the market makers of the future?

Transcript

Jonathan Sokobin: Over the course of the last day or so, there have been a couple of key themes that have been running through many of the panels. Almost regardless of the title of the panel, those themes really revolve around who's providing liquidity in the secondary market today, particularly in over-the-counter bonds, treasuries, and those markets; what's the impact of changes in behaviors of dealers, particularly dealers who are part of bank holding companies and providing that liquidity; and whether the combination of the change in liquidity in these markets, whether they're transitory or permanent, and the obligations of the Federal Reserve System as lender of last resort raised the level of questions, concerns that should be addressed by policymakers. What we are going to do the next hour and half or so is talk specifically about the question of who's providing liquidity, who are the nontraditional liquidity providers and what does it mean for the provisions of liquidity in these markets. I am joined by three experts, distinguished panelists, and you can find their bios in the book. We are going in alphabetical order to start, and so I'm going to start with Larry Harris, who's Keenan Chair in Finance at the University of Southern California. So Larry, you've got first dibs.

Larry Harris: Thank you very much. I'll get the slides up in a moment. I divided my topic, my time into two parts. The original part will actually be the second part of my talk. The first part is in response to my impressions of listening to our discussion yesterday night and today. I thought it might benefit a little bit to talk about the overall context in which we are entering this problem, discussing liquidity. But before I do that I want to start with just a very quick digression. I want to give a plug for the model that Pete Kyle presented to you earlier. If you haven't taken a close look at this model, I strongly suggest that you do so. In my opinion, it's the single most important empirical paper in market microstructure. So I strongly suggest you take a look at it.

Let me first begin with my general comments, and then we will turn to the specific comments that are relevant to this panel.

Let's talk about physical reality. What is liquidity ultimately about? We all know that someone has to own everything. Liquidity is simply about the transfer of risk from one owner, one holder to another holder. So this is really the problem we are talking about. So let's talk a little bit about risk in instruments. Instruments rarely exclusively embody a single unique risk. They generally represent many common risks that instruments share and as a consequence many instruments are highly correlated. Some of the examples are on the offer on bonds, investment-grade corporate bonds. These instruments have a degree of substitutability among them, not quite fungible but they're not quite fully independent. When we speak about liquidity, we have to be careful that we distinguish between the liquidity of instruments and the liquidity for the underlying factor risks. Ultimately, it's the risks that are being transferred. The instruments themselves just convey those risks.

And occasionally there's some frictions that make it difficult to transfer risk in one type of instrument perhaps because it's thinly traded and not very common, even though that risk represents a much wider risk that is carried by many other instruments so that's something to keep in mind.

What are our concerns about liquidity? Our concerns are pretty obvious to all of us. Extreme unexpected demands to transfer risk and destroy value—this is why we are concerned about it. What are the values we're concerned about? We think that value is lost when there's so much turmoil that organizational structures get lost through bankruptcy or when investors lose confidence and as a consequence start making decisions that are not in their advantage or to the advantage of the economies of all. And these are the types of things we are trying to preserve and this is the reason why you have a conference such as this one.

What's the role of speed? We've already talked about it a bit today and last night. Well, we're all well-organized and by that generally slow but not always, transfers of risk are usually better than fast ones and certainly ones that are chaotic. Too much too fast can overwhelm the capacity to value assets and to make decisions and the problems are people can panic and then the natural buyers and sellers—remember everything has to be held, it's just a matter of transferring from one person to another, everything has to be held—the natural buyers and sellers often can't react quickly enough to prevent excess volatility and that's the reason why we are concerned.

So what are the things that take time? It's asset evaluation, risk evaluation. and decision-making. All these things take time. If you need to do something faster than those processes can respond by the people who are going to be the ultimate natural buyers of things that other people want to sell then you run into problems. So what are the solutions? So having entered the provided general framework, we can easily now talk about what I think are the general solutions. The first is obviously to reduce all frictions, want to improve markets and reduce search cost. So ultimately if it's a matter of transferring risk from one person to another, then one person has to know about the other in order for this to happen. We want to reduce uncertainties that are caused by complexities and by lack of relevant information. So to the extent that people are unwilling to act because they are uncertain, we have to remove those uncertainties. So to this and perhaps central clearing can be very beneficial because it removes the uncertainty of the solvency of your counterparties, doesn't fully remove it, but it certainly solves the problem, and it certainly simplifies things. Overly complex instruments, of course, are not very helpful either. We've taken as given that the instruments are what they are but once potential solutions liquidity problems, simplification of the instruments so that they're easier to value and easier to transfer as a result.

The next solution is to preplan liquidity transfers using explicit options contracts. If you know that you are going to do something because that happens, then why don't you pre-plan it and pre-arrange it so that when it happens, it's already done and ready. And to the extent that people are aware of the existence of those contracts and their impact on each other, then they'll be able to plan accordingly and get a lot less uncertainty.

And then finally, and I think this is perhaps the most important thing, ensure that there are private incentives to handle liquidity in particular t I mean is ensure the liquidity is priced everywhere, so that when people put demands on liquidity, they actually feel the cost that they are imposing upon others and of course upon the transaction process. I think liquidity, especially in mutual funds, is not always priced and that we should be giving a fair amount of thought to getting it priced.

So who will supply liquidity? Now we turn to the second part and the original purpose of standing on this panel. Bond liquidity in the future, what's it going to look like? Well, clearly reducing frictions that make it difficult for the natural buyer to find the natural seller will only improve the markets in the future. If we can get the natural buyer to find the natural seller, easily, then there's just going to be a whole lot more liquidity. To this end, we have a market structure now that's only half way to where it needs to be. We have a fair amount of electronic trading, which we are going to talk about in a moment, but many people are not able to participate in those electronic systems. They have to go through dealers instead. I think dealers are going to be extremely important this market for the number of instruments that exist, but whenever the buy side can meet the sell side with less cost than going through dealers, then we're going to have more liquidity because the buy side can provide a lot of liquidity to the sell side. This is the primary reason why the equity markets, which are so much more complex, in that they have to process so much more credit risk and so much more uncertainty and yet they have so much liquidity. Order display facilities, a fancy name for essentially electronic exchanges, need to be more readily available to all traders. That by itself will vastly improve the problems that we are concerned about with respect to liquidity and the bond markets.

Finally, I think proprietary firms are going to replace traditional dealers. They already are and I'm going to talk about that in a moment. Its certainty what's happened in the equity markets and it's what going to happen in the future. The future truly is already arriving. Electronic trading and corporate bonds is growing substantially in various alternative-trading systems. It benefits the dealers and some large institutions but not most investors. Most investors are still going through dealers. Here's some facts that unless you're intimately involved in these markets you might not be aware. These come out of a paper that I imagine many are already very familiar with, that I recently wrote. It's called "Bond Transactions Cost, Trade-Through and Riskless Principle Trades," or something like that. Half the bonds on trades have two-sided quotes, almost all the time. So how do I know this? I went to Interactive brokers and they gave me their version of the national best bidder offer for bonds. They produced this because they're doing a pure agency business for their clients.

By the way, I should reveal to you that I am a director of Interactive, so take everything I say with a grain of salt but my interest in this issue long predates Interactive. Interactive didn't put me up to doing this study. I asked them for the data. They are doing an agency business in bonds and they need to know how to route their bonds so they're collecting quotes and indications but the indications are almost always firm. They are collecting them from a whole bunch of different electronic venues, and they combined it and they produced a national best bidder offer, which allowed me to ask, how often do we see bids and offers in these bonds? For more than half of the bonds in traced is a two-sided quote just about all the time. We're talking about 99 percent of the time. Two-hundred-thirty bonds trade more than 20 times a day. These look like small Nasdaq stocks, from—actually large Nasdaq stocks or medium-sized ones from 20 to 30 years ago, and the reason why is because they are trading in essentially the same market structure.

A couple of comments about dealing at banks versus prop firms. This repeats a little bit of what you heard last night and some this morning. Dealing is moving from banks to proprietary firms, and there's a couple of reasons why. Of course there's the Volcker Rule which is restricting the banks' ability to take risks. Personally I'm not much troubled by it. When they take risk using money that's been subsidized by the FDIC, I get really worried; you're all aware of the reason.

There are of course capital requirements at banks that have created an unlevel playing field vis-à,À-vis the traders at prop firms. The traders at prop firms are able to take advantage of portfolio margining. Let me tell you why this is so important. Somebody's got a 14-year off-the-run bond from an issuer and they want to sell it, and there's a 15-year on-the-run from the same issuer, so a prop trader says, I'll buy the 14-year and immediately turns around and sells the 15-year either because it's already liquid or because he can borrow it and sell it short. It's either he has it or he doesn't have it. Either way, it establishes a position that is essentially risk-free, and any decent portfolio-margining model is going to show that that's a pretty much risk-free position and it shouldn't require a lot of capital to guarantee the slight difference in credit risk and in term-structure risk between those two bonds. But if the same transaction is done in a bank, it incurs a capital charge on both legs. It's completely blind to the fact that the two sides are completely correlated. Remember I talked about providing liquidity in the underlying risk and not the instruments. Ultimately, when you have a highly fragmented market such as this one, these types of trades become extremely important, and that's why portfolio margining is so important.

Finally, the banks in many cases have simply chosen not to commit, at least in my opinion and I'm uniformed in this respects, but it seems obvious to me, that they simply chosen to commit less capital because there's less profitability because there hasn't been a whole lot of volatility. Interest rates have been pretty much stuck where they are. In my opinion, they're likely to stay there for the rest of our lives. That's a demographic issue and one worthy of a much longer discussion. Maybe you'll catch me afterwards.

Some suggestions for future improvements. I think that we ought to have an NBBO, an order handling rules for corporate bonds. That will basically bring corporate bonds with similar market structures to the equity markets. A lot of people say bonds are different than equities. From my point of view they're all securities and from Pete Kyle's point of view, there's only really a couple of things, and I agree with it, a couple of things that determine market structure and transaction cost—it's all one problem. It's a bilateral search for liquidity. And to the extent that we can make it easier for the buy side to find the sell side, so much the better. Now with the production of an NBBO that we can all trust, then we substantially improve the pricing problem for bonds because we have prices in which people are really committed to trade as a guide to the evaluation of mutual funds and so forth. I think that central clearing for repost and other instruments is really important. It will help solve the rehypothecation problem that we talked about before and in particular, in Charlie's example of a, b and c, b as a balanced position; they ought to be taken out of the story, at least to the extent that we can. It reduces the overall level of uncertainty to the question of whether a clearinghouse is solvent.

But the clearinghouse, the reason this is really needed, it's a private solution. The clearing members all have strong incentives, no, if you're not a clearing member you have a strong incentive to try to push the edge. You want to get what you can get, but if you are a clearing member, all of a sudden every member is worried about every other member, and they don't want the other guys to be pushing the edge because if they're pushing the edge then it's going to fall back on them. So collectively they have a strong incentive in self-regulation. There are problems with clearinghouses and we fear what happens if a clearinghouse fails but I think that the transparency associated with clearinghouses, the alignment of the personal incentives of the clearing members to govern each other probably overwhelm the problems that we have, the potential problems we have with clearinghouses.

Here's a couple more suggestions and I'll finish up. I think that we should require more capital for entities that make promises that are based on their balance sheets. That's not more capital for asset managers because they're not making promises that are based on their balance sheets. That's people who are dealing or otherwise anybody who's making a promise—an insurance company, that if they're making a promise on their balance sheet they ought to have capital. So more capital transfers, liquidity, and value risk to the private sector, to the equity holders of that capital. That's really good because you have to have your hand in the guillotine to pay attention. Our biggest problem when too much of that risk is governed by the government because then there's an inclination to arrange your affairs to maximize the value of the play and that's not something that we need.

Here's something that I find extremely interesting and I want to dwell a moment here before I conclude. Whenever anybody asks for more capital, bankers get really rigid and uptight because to them capital seems like it's very expensive. I've never understood this—these guys were A students in our classroom, when they studied finance and they learned about the Miller Modigliani Theorem. When you ask for more capital the cost of capital drops. So we have to ask ourselves, why the tremendous resistance to get more capital? Either there's not any capital out there, and I don't think that's the case, the government has been pumping money like you can't believe. There's capital all over the place.

So what else is going on? Maybe people are worried about free cash flow problems or maybe the whole sector just wants to push the envelope because they can increase the value of the put against the government, that's what concerns me.

And finally, I think we should get time varying market prices for mutual funds liquidity. Again, I's like to see a private solution in a particular private market for liquidity in mutual funds. Normally it's not an issue and I agree with you, Ty, that perhaps that we should just except the fact that the customers of mutual funds should get free liquidity, even though the underlying instruments are not liquid and that most of the time that's okay and that's a cost of ensuring people that this is a sensible way to invest and it gives them confidence and that's great. But I am troubled by the underlying problem that there's no such thing as free liquidity, and if we really want to ensure we don't get the liquidity problems then we have to price the liquidity.

Then the question is, how do you price liquidity? Probably the best way to do it is through some ETF-like scheme. The way I would do it for an open-ended fund is, I would say that all transaction with an open-ended fund of the traditional type have to be done at day-end as we currently are doing it, but it has to be done in-kind. Now how can people do transactions in-kind? Well, they would contract with an authorized participant and the participant would basically say "I will let you buy or sell this fund at one penny over net-asset value that the fund computes at the end of the day." And that the different authorized participants would compete with each other to provide the differential, this fee for providing the transaction. Then you can easily have an actively managed fund because then all the authorized participants just have to execute a confidentiality agreement. You don't have to have a tracking index; you don't have to have all of the other overburden that we normally associate with our nascent efforts to acquire ETFs in actively traded funds. The additional benefit of this would be that it would allow the open-ended mutual funds that we presently have to be more tax-efficient. They have a tax inefficiency relative to ETFs that will ultimately destroy that industry if people get wise to it. I'm not sure that they ever will, but it certainly should trouble people. It's another example of an unlevel playing field.

So that's it for my comments, and I really thank you for your attention.

Sokobin: Thanks, Larry. We now turn to Jamil Nazarali, who is the senior managing director and head of Citadel Execution Services for Citadel Securities.

Jamil Nazarali: My name is Jamil Nazarali and I'm head of Client Market Making in Equities, Options, and FX at Citadel Securities. Citadel Securities is one of the largest market makers in the world. We trade virtually every asset that trades electronically in every major market in the world. In U.S. equities, for example, on a typical day, we trade 14 to 15 percent of total consolidated volume. If you think about that for a second, that means one in seven shares that trade in the U.S. every day has Citadel Securities on the other side of it. In the retail market, we handle about one-third of all retail broker-dealer flow. Our clients include online firms like Ameritrade and Scottrade, private wealth management firms like Merrill Lynch and Morgan Stanley, and those in between like Fidelity.

As we think about the role of nonbank liquidity providers, I'd like to talk through two different markets that have evolved very differently, but are both entirely electronic. I'd like to contrast the evolution of the equity markets, which are largely made up of nonbank market making, and the FX market, which is also largely electronic and which nonbanks are increasing their liquidity provision, but it's still largely dominated by the banks.

So, what is it about these markets? Let me go through each one. Over the last 20 years, equity markets have moved to be dominated by the nonbank liquidity providers. If you look at retail equity market making, for example, the largest liquidity provider is Citadel Securities, with about a 35 percent market share. The number two player is KCG, with about 24 percent. Over the last 24 months you've seen all of the banks leave this space—Goldman Sachs, Credit Suisse, Merrill Lynch, and Wells Fargo, they have all exited, and the two remaining banks, Citi and UBS, have seen their market share go down from 38 percent five years ago, to 19 percent today.

So what is it about the equity market? There is really four things you need to foster the evolution of nonbank liquidity providers. The first is, the market has to be electronic. While it's not impossible for nonbanks to evolve in a manual market, it gets much harder. Nonbanks tend to be really good at pricing securities, and they tend to do well in those markets where the best price wins. That's very hard in markets that are nonelectronic. If you look at the equity markets 20 years ago, they tended to be very manual. The NYSE, for example, traded most of the securities of firms listed there, and it was an open outcry model with human traders handling orders. Across Wall Street, you had human traders which tended to trade the large blocks. Reg NMS changed all of that by requiring exchanges to route to the order that had the best price. What that forced the NYSE to do was become a fast market. Because if they continued to trade manually, then their quotes could be ignored and traded at faster markets. That was the first step in doing that but as I pointed out before, electronic markets are not enough. FX markets are largely electronic, and they are still dominated by bank providers.

What else is it? Well the second element, and Larry talked about this, is having a central market. The equity markets have a national best bidder offer, and Reg NMS facilitated that because once you have the requirement to route to the best price, you need a single market that knits together all that liquidity. You also needed to subscribe to market data from every market out there. So even though other markets existed and investors could compare those prices across markets, they didn't really exist in a central way because most people were not connected to all of those markets, and exchanges like the NYSE would routinely trade through them. Once you had this central market, it was very easy to see who had the best price and very easy to send that order to whoever did. That created the dynamic where the best price would win. So, combined with cheap computing power, cheap communications and decimalization, it really provided a very fertile ground for nonbank liquidity providers to grow their offerings.

Now, this is really different from the FX market because in the FX market there is no central market and it's very hard for someone trading on that market to compare the prices across venues. If I want to do a trade on the FX market, there are a couple public ECNs but their prices aren't that competitive. If I do a trade, I'm either trading with my bank, who I have a relationship with, or if I'm a hedge fund or money management firm, I might have some market makers that provide liquidity to me. But it's really hard for me to really compare prices across venues. On top of that, the prices are not firm. The FX market has a dynamic called last look and so the price you see may not be a firm price. When you go out to get that price, it may no longer be available and you could get declined. So, it's really hard for you to compare prices across venues.

The third thing that is really needed for nonbank liquidity providers to merge is centralized clearing. That's another thing Larry touched on. This is really, really important because for me as a nonbank liquidity provider to win, I can post the best price, but everyone has to see it, and people have to be able to trade with it. That's really hard to do if you don't have centralized clearing. So, again, the equity market's always had centralized clearing, so any broker-dealer, or anyone trading through a broker-dealer, can post a price on an exchange, and any other broker-dealer can access that price. There doesn't have to be a relationship between those two parties. You can contrast that with the FX market, where clearing is really only available to the banks. For me to trade with you, either we both have to prime with the same bank, or the bank that you prime with has to have a relationship with the bank that I have a relationship with. The four of us then have to sign a four-way agreement which is number one, very time-consuming, number two, very expensive and number three, it creates a requirement for me to have that relationship with you. I can't just go and post the best price on any venue and have any taker come and trade with me.

You really see the benefit of that centralized clearing in the interest rate swap market. Prior to centralized clearing and interest rate swaps, trading was really dominated by the Big Five banks. One reason for that is counterparty risk. Interest rate swaps tend to be very long in duration and when I do a trade, I want to do it with a party that I think is going to be around for a long time. That really favored the top five banks. Once you instituted centralized clearing, you enabled nonbank liquidity providers to emerge. Today, in just a very short period of time, Citadel Securities for example, has become one of the top three liquidity providers in interest rate swaps. That would never have happened without centralized clearing. To summarize here, as Larry pointed out and as we all know, nonbanks are providing less liquidity in the market partly because of less risk taking and partly because of more stringent requirements around capital. That provides an opening for nonbanks. There are really a number of different factors that are required for the nonbanks to provide that liquidity including electronic trading, a centralized market place where you can compare prices across venues, and centralized clearing.

Thank you.

Sokobin: Thank you, Jamil. Finally, I'd like to introduce Barbara Novick, vice chairman, BlackRock Inc.

Novick: I'm going to assume that most of you have heard of BlackRock and you probably know we manage about $4.5 trillion, but it's other people's money. Last year, I spoke here about a framework for regulation in the asset management industry. In laying out that rationale for what I called the products-and-activities approach, I touched on the importance of liquidity risk management, especially in mutual funds, but truly across all of the portfolios that we manage. We've written quite a few papers, and there are two in particular that you might find interesting. Both are called, "Addressing Market Liquidity." Think of it as one and two, and you can find more detail about today's conversation and remarks in those papers.

This year I am going to focus much more on liquidity in what I call the broader fixed-income ecosystem. I will try and tie back some of the liquidity risk management and market liquidity to the questions for this panel. We heard last night that there are multiple definitions of liquidity. These include market liquidity, funding liquidity, and a whole lot more. I think if we just did a survey, I would have loved to have the capacity to ask questions. If we did a survey and asked everybody in here to define liquidity, you'd be shocked at how many different definitions people have for that one word. That leaves to part of this "Tower of Babel" around liquidity. Unfortunately, most discussions about market liquidity very quickly morph into what is called fund redemption risk. I can't emphasize enough that market liquidity and fund redemption risk are two really different things. I'm going to try and demonstrate that.

The bond markets, of course, have been evolving. They have been evolving for as long as I've been in this business, which is, I'm sad to say, more than 30 years. I started when I was 12. And of course, asset managers and other market participants keep adapting to those changes. So, for a panel about liquidity and a conference about liquidity, we'll just say it's a fluid situation.

And some of the things that we do in terms of evolving, how we build our portfolios... My friend this morning talked a little bit about some of the things they are doing in liquidity risk management. But building the portfolios...some of the securities you buy, you might be buying them with the intention of having a longer holding period and striating your portfolio, knowing that you need more liquidity from some parts than other parts. Holding more liquid securities... We see in high-yield funds, we did a study in one of our papers where we looked over time at what were the cash levels in high-yield funds. They are not static. For portfolio managers the first line of defense are very actively thinking about liquidity risk management, different portfolios, different things, different points in time, different answers. You can't look at it as a static question. Of course how we trade the bonds, and a lot of this panel is about breaking trades into smaller lots. Using electronic mechanisms...whether it's all to all, or dealer, or any kind of combination, we've been very much pounding the drum for several years. We need more of that. We need to move from a principle-based market, to an agency-based market, and that's one way. We also need to change behavior. At BlackRock, our traders are trained not to just be price takers, but actually to be price makers. By posting what they'll take, at what price, or where they'll sell something, they become a provider of liquidity as opposed to as just a taker of liquidity.

We've also changed over the years how we monitor and how we assess liquidity risk. Having been at BlackRock from the beginning I can say that in our first marketing documents we talked about five sources of risk and fixed income...duration, convexity, credit, foreign exchange, and liquidity. It's still the same five risks, nothing has changed except the order of them. Today, in this market, liquidity probably moves up in that list and so that's something we have to factor in. We've also very publicly endorsed the SEC's proposals. We think that there's a need for raising the bar on liquidity risk management. We also think there's a need for expanding that tool kit of options. So, I'm not going to argue with one paper or another paper. But clearly, there is an opportunity to take the best of all from around the world and put them together. Swing pricing...most fund managers use that in Europe today. Let's figure out operationally how to make that work in the U.S. Let's use it in the U.S. It's investor friendly and systemically risk friendly. Or redemptions in kind—it wouldn't work for individual investors but it certainly would work for institutional investors. It's already used in many fund complexes. Why not make it a standard operating procedure that if you have a large order for a redemption, you can give them securities. It's not off the charts, it's used in many fund complexes quite often. So, there are tools in that tool kit...some already there...some that can be developed...some that can be enhanced...and these are partial answers.

Let's take a moment and turn to some of the data. Hopefully, I'll open your eyes to some things that are surprising. There are four facts about liquidity that everyone's going to agree on. First is that broker-dealer inventories have declined. It doesn't matter why, it just is. Second is bond turnover has declined, but I'm going to come back to that one. Third is corporate bond issuance is at record levels. And fourth is bond funds have grown significantly. We can all agree on those. Anybody want to take the other side? [No response.] OK, good.

While all four of those statements are totally true, it's a partial picture and it's incredibly misleading. Let me show you how and why. This morning Greg mentioned that the bond turnover is down 80 percent and called that alarming. I thought it was alarming, too. So, I started trying to figure out how do I reconcile corporate bond issuance at record levels...everyone saying that turnover is low...there's no liquidity. Doesn't make any sense. So, I started looking into the numbers. If I told you that turnover is a ratio of the number of transactions divided by the outstanding bonds, and if I told you that both the numerator and denominator have grown since the crisis, you'd probably be pretty surprised. But that's what the data shows. Yes, the ratio has declined, but it's declined because the denominator has grown more rapidly than the numerator.

Now, go back to what the denominator is—it's corporate bond issuance. So, if you're a treasurer in the crisis, you get caught with your proverbial pants down, you have too much commercial paper, you need to roll it, you can't. You come out of the crisis and you say, "Hey, I'm going to extend that." So you now have another source of new issuance. You have somebody else who says, "I want to do a stock buyback. Let me issue corporate bonds. Let me be opportunistic. Rates are at ridiculously low levels. Let me issue corporate bonds." So, the denominator has gone up so quickly that that turnover ratio is down. But that doesn't mean that there's no transactions. You just saw from one of my co-panelists that, in fact, there's a bid-ask on many of those bonds and there's a lot of turnover on some of those bonds. So, take the numbers a little bit in stride and recognize we have to go deeper. Same thing on bond funds.

If I told you that there's over five trillion dollars of corporate bonds held in mutual funds, you'd say, well, that's a huge increase over time. If I then said, well, how much of those are held in dedicated bond funds? Well, the answer is 3.5 trillion. Where's the missing 1.5 trillion go? Well, there's things called target date funds, balanced funds, asset allocation funds, and many different multi-asset class funds. Well, that's 1.5 trillion out of just over 5 trillion. Now you take the 3.5 trillion. Morningstar has 46 categories of bond funds. So, we're not talking about a homogenous data set. We're talking about lots of little slices. Frankly, I don't think I can find anyone in this room who would say an intermediate bond fund that includes treasuries, mortgages, and corporate bonds is going to have a run that they can't meet their redemptions. It's just so unfathomable beyond any stretch of anyone's imagination.

So, when people talk about corporate bond funds, what they're really talking about is, a tiny slice of this whole universe. Because there are almost no investment-grade corporate bond funds, they are talking only about high-yield-only funds. We've seen data from December, which was the perfect storm. Third Avenue closed their fund somewhat unexpectedly, and I'll say, a little bit sloppy. We also had the Fed raising rates and everyone thought, oh, my, chew our fingernails off. Come in Monday. What's going to happen? The world's going to end. But in fact the world didn't end because the other thing that is never discussed is what I call the other asset owners. So if you look at bonds—even if you took all the bond funds, even if you took all the bonds that are in multi-asset class funds, you're talking about less than 20 percent. So someone else owns 80 percent. Let's study that for a change. Do you know what happened in December? There were huge outflows we saw earlier from high-yield funds. Some of it went from one fund to another fund in the same sector. Some of it was insurance companies and other investors coming in and saying, these are cheap.

You also have to look at the overall environment. High yield has a huge component of oil and other commodity bonds. What was happening to the equities in those companies? They were going down. So, of course those funds were going down in price. How could they not? So you have to be looking at that whole picture and that's why I said looking at the partial data. It's true, but it's very misleading.

I know one of the most popular questions before was bond ETFs but we never really got an answer or a commentary on it. So I want to give you some intriguing information on bond ETFs. Although they are probably the most maligned security or fund out there, you'll be surprised to hear me say that they are actually a part of the solution. This whole question of where will liquidity come from. It turns out that bond ETFs do not redeem in cash. The investors cannot come to the fund and say, I want money and you have to sell securities. That's a complete fantasy. The way they work is very simple—redemption in kind. So if an investor wants to get out, they sell their securities in the secondary market, and if there's enough of an arbitrage between the price of the fund and the price of the secondary, an approved provider will come in with a basket. Either they will take a basket or they will give a basket. So there's no concept for redemptions for cash and that secondary market trading provides enormous amount of liquidity. Where have we seen it? And the data is quite compelling. Whether you look at Bill Gross leaving PIMCO, you look at Third Avenue, what you will see each time is, when bond markets are under stress, there is a spike in the ETFs secondary market trading. And that is acting as, essentially, a cushion or a shock absorber, for people who want to make or remove a capital bet in that sector. It's very interesting because the price of going in and out of ETFs is a stock market price.

So, you've essentially gone from that dealer balance sheet, principal trading, to an agency market, a stock exchange—the ultimate.

Of course, I'm going to add to that and just say that if you look at the trace data and you talk about turnover, keep in mind bond ETF data, because it trades on a stock exchange, is zero in trace. It's not reported. So, the trace numbers, in addition to having a denominator that's rising much faster than reality, a different interest rate environment, I assume we wouldn't see quite so much issuance. In addition, the numerator is artificially lower. In addition to the bond ETF's, why else is that numerator lower? Well, it turns out bond funds have been in inflows for the last several years. So if I'm a portfolio manager and I have the idea that I want to change my asset allocation within that portfolio, I have choices. I can sell things and buy other ones, or if I am in cash flow positive mode, I can choose to deploy the new cash to where I want to reweight the portfolio. I don't have to sell anything. So, we have a period of time where the inflows to bond funds have been so strong, there's not a lot of reasons to go sell the ones you have unless you have a strong view on that individual credit. When you look at that, you look at insurance companies, you look at defined-contribution plans, the growth of target date funds—you have so many structural changes in this market that affect the real turnover, the transaction volume, that you can't really compare precrisis to today or 10 years ago to today, you have to be looking to what I call the modern markets.

Now I'll come back to some of the questions that were posed. Should we be concerned about changes in liquidity and who will be the market makers of the future? As I disclosed earlier, the risks are the same today as they were, but that doesn't mean we should ignore market challenges. We have to evolve. It includes evolving and looking at the market level as well as at the fund level, and the only thing I can say is constant is change. So, the capacity of broker-dealers to conduct the market making—it's restrained, it's different, it's changed. We accept that and move on. Market participants are beginning to supplement that. They're doing more agency-like activities, whether it's electronic trading or it's the growth and institutional use of bond ETFs. Likewise, we see at the fund level, portfolio construction, the increased focus on liquidity risk management, becoming price makers—all these things come together.

So, I'm going to end by answering the questions which are basically, yes we should be, and we are concerned about changes in liquidity. But before we bemoan the loss of liquidity, we need to look at that whole picture and encourage everyone to look beyond the headline numbers. We can see even from this panel there's an evolution, a sea change of where the market making is coming from. As I said earlier, it's a very fluid situation. Thank you.

Sokobin: Thank you. This has given us a lot of food for thought. I want to start with a couple of questions. The first is I want to refer to an interesting piece put out by the Bank for International Settlements recently on the growth of electronification of debt markets. They posited that there was a difference in the way that, let's call it traditional market makers or traditional liquidity providers provide liquidity versus nontraditional, nontraditional being represented on this stage. What they said was, a traditional market maker is going to, in the face of some uncertainty, widen spreads but maintain depth. And a nontraditional market maker is going to maintain tighter bid-ask spreads—lower spread—but then reduce the amount that they are willing to transact. One, I'd like you to comment if that's what you see, and two, it could be one of the reasons why the measures of market liquidity that regulators see don't sort of give a meddled picture. So, the question ultimately is, when you have different kinds or different models of the provision of liquidity, what does it mean for the stability of the liquidity in the market? How do those two models interact with each other, and what does that mean for a regulator?

Nazarali: I'm not sure about this phenomenon of tighter bid-ask spreads versus less depth. I would guess that it's probably less of a factor but I'd have to look at the data. I think one of the elements of electronic trading is that things happen very, very rapidly. When markets are manual, when there's a shock to the market, it make take a while to happen and then it may take a while to recover from that. For example, in the 1987 stock market crash, the market was very manual, so when the market crashed, number one, there was a lot more illiquidity because you couldn't get your broker on the phone to execute a trade. If you contrast that to the flash crash in May 2010—when the market crashed then, you could always get your tradeoff if you really wanted to.

The second thing is that the market reacted very quickly when it realized that it wasn't caused by some other fundamental driver. You saw the same thing in the treasury market a couple years ago in January, where you had a little bit of a flash crash, but also a very rapid recovery. I'm not sure about this depth of the market versus other, but I do know one aspect of electronic trading is that a big liquidity event can have a bigger impact. That's because firms will react very quickly to it, but there's also a much quicker recovery to that. All during that time there tends to be a lot of liquidity, so you can get your trades done and you don't have this inability to trade.

Novick: Two things I would add: one is, I think we have to think of it conceptually like the decimalization of equities. There is a structural, fundamental change going on in how securities trade. We do need to go from the need for a balance sheet, to an agency like execution. That's just, I think at this point, a given, given where regulation has taken us.

Harris: I actually think there are three models we should be talking about. The traditional OTC [over-the-counter] model is a model where the dealer needs to get the customer to call them in order to do the deal. When that's the case, the dealer has to meet the needs of the customer. If a particular customer's got a volume need, I would expect to see the capacity to trade remains relatively constant while they widen the spreads if they're concerned. The other extreme is a pure agency model or anonymous trading. When people get nervous, they're just not going to show as much size because they want to be nimble and get out of the way. Orders get broken into smaller and smaller pieces. The fact that we see that in fixed income the trades are smaller just means that people are changing the style of trading. There has been motion from the first type of market to the other type of market.

The third market structure that we probably should mention for completeness is the inter-dealer brokered market. In the inter-dealer brokered market, we see it mostly in treasuries but not so much as in corporate fixed income. What you've got is a system where the dealers are trading with other and are very reluctant to show their complete size to each other. So, what they do is they take turns showing size as they build up trades. There the story is that size is proprietary information. They don't want other people to know what they're doing. That dynamic preserves the option to decide just how much you're going to build up. To some extent, we'll see that when there's more uncertainty, the people aren't going to show as much. What I find interesting about that market is quite often a broker will build a trade between dealer A and dealer B, and at some point dealer A or dealer B doesn't want to show any more size and the trade is fixed and it's then finished. Then a moment later, the broker's there brokering the same trade between A and B, but A doesn't know he's now trading again with B, or B with A, simply because one of the two parties decided that they didn't want to show as much to the other. There's some dynamic going on there as well. So as to how the BIS [Bank of International Settlements] got their results, I don't know...I'm not familiar with it, but it doesn't surprise me about what I've heard.

Sokobin: I'm going to pick on you for a second, Barbara, and ask a question. You've stated this notion that balance sheet has moved from traditional brokers, banks, to asset managers. Help us understand the conditions under which an asset manager may be willing to provide short-term liquidity, and when they might not.

Novick: OK, let me clarify. An asset manager is never providing liquidity as a balance sheet. But as a portfolio manager, you can choose to be what we call a price maker instead of a price taker. In the normal model, you would wait for something to be shown to you or you would say I'm requesting a bid for. In today's world you might say, I'm willing to buy X amount at this price, or I have this much to offer at a certain price. So, it changes the dynamics. Rather than you taking liquidity out of the market, you are now offering liquidity because you are showing where you would take something. That's not the same as short-term trading on our own balance sheet. We absolutely don't do that.

Sokobin: Do you think that model remains the same in periods of higher volatility? Do you think that you act differently in that kind of situation?

Novick: I think what we've seen in periods of market stress is asset managers of all different kinds—in-house managers, insurance companies, sovereign wealth funds, pension plans, outside asset managers, all sorts of asset managers—use those things as opportunities to, in the case of bonds, buy bonds cheap. We saw that explicitly in December. There were clients who came to us and gave us new cash to invest in high yield because they felt it was so low, compared to what they thought was sort of intrinsic or fundamental value, that they wanted more exposure in that sector. So that puts you, as a manager, in a different place. You now have cash to deploy into that sector and that would be true whether someone was managing their portfolio themselves, or if they were having an asset manager do it for them.

Sokobin: Do either of you gentlemen want to add to this conversation?

Harris: Sure, I just observed that Dimensional Fund Advisors is a very large institutional indexer in the small cap equity space. For many years they were able to produce a return in excess of their index bench mark because they were supplying liquidity on a regular basis. They would wait until people would come to them with blocks saying, would you like to buy this block? And they'll say well, when you've exhausted all of your other alternatives, come back to me. Often those people did come back and because Dimensional was regularly growing due to receiving new capital, they would be buying these blocks at discounts and effectively providing liquidity to the rest of the market, even though they were essentially an indexer. So, adding value through their trading process, I wouldn't call them an active manager but they are active at least with respect to liquidity. This is just another example of buy-side meeting buy-side needs.

Nazarali: One of the interesting things here is that, in any market when you have the withdrawal of certain players, it creates opportunities for the other players. So as banks withdraw their liquidity provision, that often manifests itself in wider bid-ask spreads, which creates a profit opportunity for nonbanks to come in there. So as spreads widen, we look at that opportunity and say, OK, we can actually make money off of that and that does provide a big incentive to come in there. But as I was speaking about, there are other factors, including having a centralized market place and centralized clearing, that really facilitate the large migration from banks to nonbanks. So, it provides an opportunity but it's not sufficient in and of itself.

Sokobin: I want to weave in some of the questions that have come in through Pigeonhole and I'm going to start with first, which is about PTFs. So, still on this question of who's providing liquidity and what are the implications? The questions are, are PTFs helping liquidity or hampering it, and how?

Nazarali: Well, they're absolutely helping liquidity. People are trading with them because they're at the inside market and because they are providing tight prices. No one's mandating the emergence of high-frequency trading. No one is saying, hey, you guys, we have to trade with you. What's happening is that they're coming in there and providing a lot of liquidity, and that's why they're emerging. I think that there's no question that they're helping liquidity. You see this much more in the equity markets because it's much more developed and you've had automated trading existing there for much longer. There have been venues that have said, look, HFT [high-frequency trading] is bad. I'm going to protect you from HFT. I'm going to create an alternative trading system which has no HFT. When that happens, invariably there's no liquidity there. Because the moment in time when asset manager A wants to buy, it's very, very rare that asset manager B wants to sell at that exact moment in time. So, what we do is provide the liquidity to bridge that temporal gap, and without automated firms, there tends to be very, very little liquidity. So every one of those alternative trading systems that have gone on to flourish, have later gone on to then embrace automated market makers. I would say unequivocally, they are adding liquidity to the market.

Sokobin: In the equity space, at least the evidence that's there, and granted it's on a fairly limited set of data, seems to suggest that the PTFs are adding liquidity in the more already-liquid names and acting more as a fundamental trader in the smaller, less active names. There is starting to be some evidence in the electronic bond markets that electronification is adding liquidity to already liquid names and sort of worsening liquidity to less liquid names. How should we think about those facts in the context of what you are telling us?

Nazarali: I think the first part of that is true, that automated market makers are providing more liquidity in those instruments that are highly liquid and electronically traded. That does help the market. We've seen that when there is a withdrawal. For example, after the financial crisis in September of 2011, there was an SEC ban on short sales of financial institutions. That had the impact of making it very hard for automated market makers to do what they did because you can't buy and sell something if you're restricted from short selling it. So, you immediately saw automated market makers withdraw from the market, you saw bid-ask spreads go wider, it cost much more to trade, and volatility went up. Even those very liquid names without the impact of automated market makers suffer. Yes, it's true that there is more of that liquidity provision in the liquid names, but that's actually really important. I think that in the less liquid names, I would disagree that, in the bond market, that it hurts liquidity. I would just say there's less liquidity provision because most automated market makers don't have huge balance sheets. If something doesn't trade all that often, you're not going to warehouse it for several weeks and you just provide less liquidity. But, you do still provide some liquidity.

Novick: I would like to add to that. In the last 30 years, the one thing that has been constant is there are bonds that are more liquid and they tend to be the larger, sort of like an on-the-run concept. And there are bonds that are less liquid and they tend to be the smaller issues. We used to say that some of the new issues, after they have seasoned, they went to bond heaven. What's bond heaven? Bond heaven is the portfolios that don't turn over. The average turnover in an insurance company portfolio, let's say, is very, very low. They're going to buy those, they're going to hold those. They have to hold fixed income as a huge percentage of their portfolio. So, unless they decide that they don't like a name or they see something that is actively better, they're very unlikely to trade that bond because most of them are taxable. They have tax consequences of that trading. So, the reluctance to turn the portfolio over is very, very high. A lot of those bonds, a lot of the new issues, they start getting seasoned. They go away and the focus on new issues has been part of the market for the last 30 years. I mean, people are talking about it now like it's new? It's just not. There's also what's in the index. So, if you take, let's say, the Barclay's aggregate versus what's not in the index, and that generally is also size-related. So an individual issuer has a choice. They can issue 50 different bonds over time and have lots of little bonds outstanding, in which case they're bonds are generally illiquid, or that same issuer can choose to do fewer, less frequent larger bond offerings, in which case they're bonds will be in the index and they'll trade better. It's almost like a stock. You can choose to issue certain ways and have more stock analysts follow you, or you can choose to be illiquid and not do it that way. So, some of this is in the control of the issuers to make their bonds more attractive if secondary trading is important to them. And for many buyers, turnover just isn't a big issue and bond heaven is there for a reason.

Sokobin: You know with more and more of the trading being focused in bonds that are 90 days or newer, bond heaven is getting to be a very crowded place.

Harris: Despite the horror stories that are relayed in Michael Lewis's book Flash Boys, most high-frequency trading is just standard dealing—connecting buyers and sellers who arrive at different points in time, and standard arbitrage, which is connecting buyers and sellers in correlated assets who arrive at different points in space at the same time.

It's useful to understand why the high-frequency traders have displaced the traditional traders and the reasons are really quite simple. Once the systems are built, the computers only operate on electrons, which are pretty cheap. They have incredible reaction times; they're extraordinarily fast. They have incredible scope of attention; they can watch many, many things all at once. They follow the rules that they're programmed to follow exactly, so they don't wander off because they're inattentive or imaginative or who knows what. On top of that, they don't demand raises; they don't drop crumbs into their keyboards; they don't argue about the sizes of their offices. The bottom line is that they are just economically more efficient. It's a more efficient way of doing business. Not surprisingly, that efficiency has displaced those folks who are doing essentially the same business using technology that is just more expensive. So, who's the beneficiary? To some extent the benefits are captured by the high-frequency traders because they are able to do better, what other people couldn't do so well. But they compete with each other and as long as there's enough of them competing with each other, ultimately the benefits accrue to the buy-side through greater efficiencies. So overall, the high frequency trading doesn't worry me. What I worry about is that because of the extraordinary cost of acquiring high-speed trading systems, there are barriers to entry that could potentially be problematic. One final comment. Why are they so fast? Not because they need to be fast. Nobody makes a capital decision in a millisecond or a microsecond. The reason they need to be fast is because they have to be faster than their competitors. They need to take an opportunity when they see it, they need to be first in line when they want to be, and they want to cancel when they want to get out. Only the guy who is able to do that faster than his competitor, is the one who ultimately survives. Speed by itself doesn't matter. For them it matters because they just want to be faster than their competitors. So there's something of an arms race that should concern us all, but that's not really our discussion today.

Sokobin: Notwithstanding your comment, Larry, about the long-term expectation of interest rates, the audience is very interested in the panel's prediction of what happens to liquidity in this market and how it's being provided under different conditions. There was one question about what happens when volatility rises, interest rates rising. In the extreme is what happens if we should find ourselves again in a moment of significant disruption to these markets? I'd love to hear from the panel.

Novick: I think we should start looking at that question. What things had a bid and what things didn't have a bid in the crisis? So, subprime did not have a bid and that was very simply, you had something that was triple A—blink—we don't know if it's worth anything. There was something very, very broken with securitized assets. Only someone with a huge risk appetite would take a chance on buying something that they had no way of valuing, or having any idea what it was going to be worth in a day or a week. That's very different. If you go back at other points in time... You know we had a credit crunch, I think it was '94. We had something else in '98. There have been episodes throughout time where, in those liquidity or credit issues, people who had risk-taking appetites or who had cash that they could deploy, cleaned up. [Unintelligible] widened out hundreds of basis points even though there was a problem in Russian bonds.

People like to say that the next crisis will be different than the last. I certainly hope so because I hope we're not going to have a subprime repeat. Hopefully we've learned our lesson. But the things that were not able to be valued in the crisis—I think last night Bill talked about three weeks to get a price—that's clearly something that you have no idea what it's worth. That's very different than saying, what are the IBM bonds? Or, what are the General Electric bonds? Or things that you can least have some idea of value. You might bid it back, but that's not the same as saying there's no bid.

Harris: When a large fraction of the holders of an asset or of an asset class lose confidence in that asset, and when potential holders have likewise lost confidence, the only thing that's going to happen is that that price is going to drop. There's no amount of compelling dealers or anything else to hold the price up. So our object is not to figure out what to do if that were to happen, our object is to figure out how to keep it from happening. Confidence is of course very important and confidence is based on familiarity. As you know, you can only be confident with a relationship with a person, a machine, or an instrument if you're familiar with it. So it's very important that we create an environment where people can be confident because they have the information to become familiar with what they are dealing with. We can promote confidence by trying to reduce uncertainty in our economy, and by ensuring that there's enough information so that people are able to value things and make good decisions.

Nazarali: I agree with all of that. In terms of financial crisis, you have a mismatch between buyers and sellers, and the nature of the liquidity providers is not going to change that. So whether it's electronically provided by automated market makers or by people sitting in the bank, it's not going to make any difference. The price is going to be the price, and it's going to be hard to sell. That's just going to be the nature of it, and it doesn't matter who is providing that liquidity.

Novick: If I could add one thing on confidence. I couldn't agree more. If you really think about the crisis—so money market funds, when the Reserve broke the buck, other money market funds did not have the same problem as the Reserve, and yet you saw lots of withdrawals because people had lost confidence in the market. The market was already illiquid before Reserve broke the buck. They didn't take their money and put it in their mattress. If you look at the flows... The SEC did a study. Money came out of prime funds and went into government funds. When the insurance program was announced, money came out of government funds and went into prime funds. So confidence is a very important factor. In the theme of, the next crisis will be different than the last crisis. I think one of the things that we're not focusing enough on right now is what I call bond holder rights. So if you look at the City of Detroit and their bankruptcy, the bankruptcy judge put the pension holders above the bond holders because the pension holders were so destitute. I can understand from a human perspective, but the uncertainty that that creates for bond holders. People talk about the cocoas. After Novo Banco, the Portuguese central bank took the bonds from one bank, a good bank, and put them in a bad bank. The bond holders in those asset classes that were taken said, wait a second, we're crying foul. Why would you take pari passu claims and make one good and one bad? So, I think that this whole bondholder rights theme is extremely important to investor confidence and, in turn, to market liquidity. Because, if you saw what happened in the aftermath of Portugal, suddenly Italian bonds didn't trade so well, and other bonds didn't trade so well, and it looked like there was going to be this contagion and people started talking about cocoas. As an investor, do you have certainty? Confidence, predictability, transparency to what's happening is extremely important. I would actually put that on one of the highest rungs of the stability conversations.

Sokobin: I'm going to turn the topic for a moment. Some people have been interested, and one of the underlying themes here has been electronification of debt markets. Let's take a look at this question. What role do you believe, when you look at fixed income markets in light of Facebook, Uber, Airbnb and all, that digitalization and using networks have to find liquidity going forward? Larry, you spoke about that a little bit in you opening remarks. Would you like to start here?

Harris: So these entities—Facebook, Uber, Airbnb—are characterized by systems that allow, basically, buy-side to meet the buy-side directly. They're not going through intermediaries. These are essentially brokerage systems that allow people to meet each other. In the fixed income markets, we don't quite have that yet. One of the reasons we don't have it, especially for the smaller invertors like retail, or the smaller and mid-size institutional, is because when they go to a broker and if the broker isn't busy selling them the bond at an egregious markup, they're selling the order to a dealer who will fill the order at a high markup. So the system doesn't demand those brokers to participate in a wider, agency-based system. So you've got something of an agency problem in the payment of order flow, in which the brokers are acting in their self interests, it's not always in the interests of the customers. You don't see that kind of stuff going on with Uber or Airbnb. Facebook, of course, is a little different than those two. The problem is that we somehow have to figure out how to get brokers to give better access to these markets. Now, some do of course. Interactive brokers give their clients agency access to these electronic trading systems and I believe that Raymond James, or one of the James brokerages, does as well. There are probably others that will as well. As the customers demand these services, we're going to see more and more of it and we may see a fairly quick evolution to a more agency-based model. The SEC could be very helpful in this respect.

What about the MBBO [MIAX best bid or offer]? If the SEC doesn't ask for, or demand an MBBO from the industry, will it be produced and disseminated to everybody? Well, [unintelligible] is doing it, but they're all basically doing it for their clients; they're not selling it. What if they went out and tried to sell it? I venture that if they did, and I have no specific knowledge, the people supplying those quotes to them would say, hang on a second, I'm not supplying the quotes to you because, frankly, I don't like that product to be in the hands of everybody. The knowledge of what trade prices could be is extraordinarily powerful. Here's a result from my study about trade throughs. It turns out that about 40 to 50 percent of all trades that are reported on TRACE [Trade Reporting and Compliance Engine] are trading through a standing quote, or indication, on one of these electronic systems. And those trade throughs are really expensive to the people who... So if you have a buyer who's trading through a standing quote—maybe the quote is at 101 and the trade takes place at 102—the buyer is wondering why did I pay 102 when I could've paid 101? And the seller at 101 is saying, what's going on here? I'm offering 101, why did I just see it trade at 102? The dealers that are making money from the fact that these markets are opaque, they don't want to share this information. So, I don't see that we're going to see an MBBO created without the efforts of regulators stepping in. I don't like asking regulators to do much because I feel that they'll do too much or they'll be too stupid or something. But we have plenty of precedent with respect to the structure of markets in general. We know how equities trade. We know that equities that look like bonds trade pretty well.

Here's a fact that will just blow your mind if you're not already familiar with it. We used to have exchange trading of bonds almost exclusively at the New York Stock Exchange in corporates through the 1940s, and it continued to trade on through the New York Stock Exchange system, and in munis through the '20s. So, [unintelligible] and Rick...I just lost his second name. Anyway, these two authors looked at the transaction costs in an exchange traded system, New York Stock Exchange, from a century ago and they found that the transaction costs then were cheaper than they are now. This was before we had computers and before we had databases. They had a cabinet called the bond cabinet and that's where they put the orders. They had no electronic order routing systems, they just had a matching system that worked to match buyers to sellers on an agency basis. So things could be a lot better, and they will be.

Novick: I guess I would say, what those three companies have in common is innovation, and I believe that when there is a profit opportunity, there is innovation. We don't know exactly what format it's going to take, we don't know exactly what it's going to look like, but I think it's safe to say that if we're sitting here in five years—and, Dennis, I hope you'll invite me back—we're going to see some innovations in this area, and they're going to be game changers.

Sokobin: So, before you answer, let me add one other piece to it: there is an observation that even though we see more electronification in this market, it's really two different markets that are growing side by side. So almost all of the activity that we see in dealer-to-customer electronic interaction is in an RFQ [request for quote] system. So it's effectively an electronic version of calling dealers. Right? Whereas the dealer-to-dealer market is much more moving towards something that looks like an order book that can be viewed—you know, click to trade, all that kind of stuff.

So, how should we think about the evolution in this market if it's growing in these two ways in parallel? I don't know if that throws you off.

Nazarali: No, I mean, OK. So I'll take both of those. But if you look at those companies, really what Uber and Airbnb do is they reduce the search cost and allow someone providing a service to connect with someone who wants to be a consumer of that service. So that's very applicable to finding a buyer and seller.

What I think that, in the fixed-income markets, those—and the financial markets—those types of things already exist, so you have speed for dealers. And so I think that you already have those networks, but the real problem is you don't have the national market, to Larry's point, until you create that national market that really knits together all of those private networks, that's not going to happen, where you have the easing of the liquidity provision.

Harris: So, the key is getting the buy side to meet the buy side, and the benefit comes twofold: one benefit comes from a reduction in search costs, and the reduction in search costs more efficiency and there's more liquidity.

But there's a second benefit that's far more subtle, that's very important that I want to share with you. So let's talk about how dealers respond to better informed traders. So when trading with better-informed traders, dealers often find themselves on the wrong side of the trade. They end up buying when the better-informed traders are selling, and then they end up losing because the price is wrong.

And their response to this—because they don't have to be in the market—their response is that they tend to widen spreads so they can recover from all traders what they lose to the better-informed traders. And they have to do this because they stay in business, they're in competition, and ultimately their competition determines how wide the bid-ask spreads are going to be. So the spreads will reflect their cost of doing business, the normal cost of doing business, but also in addition, the cost of trading with informed traders. So, dealers tend to widen spreads in response to the danger of trading with informed traders.

Now, what about buy-side traders? A buy-side trader often comes to the market committed to doing a trade, and they have a choice to take liquidity or to offer liquidity. If they offer liquidity, they have a hope of getting a better price, which is a good thing. But now, how does the buy-side trader respond to living in an environment where there may be better-informed traders? So imagine that the buy-side trader is a buyer, OK, and they have two alternatives. They can buy immediately using a marketable order or they can offer liquidity. If there are informed traders present, and the informed traders are sellers—if the buy-side trader buys with a market order, he's going to lose because prices are going to drop if the informed trader indeed is better informed. Because the informed trader is selling.

If the buy-side trader says, "I'll offer liquidity, and I use a limit order," he's going to trade with somebody, probably—possibly—the informed trader, and he's still going to lose money. OK? So he loses either way.

But now, let's flip it around just a little bit and ask ourselves what happens if the buy-side trader comes to the market, wants to buy, and there are informed traders present and the informed traders are also buyers, so that the informed traders think that prices are going to be higher. So now the buy-side trader has two choices: use a market order or use a limit order to provide liquidity. Now, here's the problem: if they use a limit order and they fail to trade—this is a trade that they were precommitted to doing—they're going to lose the opportunity to profit from a trade that will have proof to be profitable.

If they use the market order, of course, they'll profit.

So what's the response of a buy-side trader who is committed to doing trading and who lives in a market where there's a possibility of trading with better-informed people? Their fear is not that they'll trade with the better-informed person and be on the wrong side of the market, they're going to be on the wrong side of the market anyway if that happens. Their fear is that they're going to fail to trade when they wished they had wanted to trade. And so how do they respond to that? Well, either they take liquidity using a market order but if they still want to offer liquidity, what they do is they become more aggressive. And so the precommitted trader's response to what economists call "adverse selection," the asymmetric distribution of information, the idea that there are people out there that are better informed than you are.

The precommitted trader's response is to become more aggressive and to narrow the spread, whereas the dealer's response is to become less aggressive and widen the spread because they have to survive. And the difference has to do with the difference between being a passive trader hoping to profit from bouncing back and forth, and an active trader who is hoping to...who's precommitted, hoping to get their trade done.

And so, what this means is that buy-side traders will displace dealers, because they can narrow spreads to the point where dealers won't profit. But that's OK because they're willing to offer better prices than the dealers are. And so we get two benefits from the buy side being able to meet the buy side, we get the benefits of a more efficient system because we've lowered search costs because the only way it can happen is through a better system.

But we also get the benefit of a different response to well-informed traders. And what does this do to the well-informed traders? It makes the world even better for them, because the spreads are narrower. There will be some losses, but that's just the nature of the world.

Sokobin: Was there something you wanted to add? OK. I'm going to push back on one last thing here. As I think about Jamil's comments about the things necessary to have pools of liquidity, some of it actually has to do with asset design. Right? So the only way that you're able to attract enough liquidity—because liquidity's an externality, is if there's enough similarity between the assets. In the extreme, it's the argument that you can't, we'll never have a wholesale take-up of electronic trading of corporate bonds because the assets are too idiosyncratic, too heterogeneous.

Larry's comments about all-to-all, or the ability of buy side providing intermediation or liquidity really is, it comes down to how wide their portfolio is, how close assets are in substitutability, all of those kinds of things. So I'm going to come back to where we started: if the issue that we're interested in is in provision of liquidity, as it relates ultimately to stability of the system, what are your thoughts on the limits of the ability of nontraditional liquidity providers to fully fill this space that was provided by traditional dealers in a world where there's, in assets where there is a lot of heterogeneity versus ones where there aren't.

Nazarali: So, yes, there is a lot of heterogeneity in the bond market, but I don't think that in and of itself that precludes nonbank [unintelligible] from playing major roles. So, for example, if you look at the listed equity options market, that's a great example of a market where there's a ton of heterogeneity, and where there is literally thousands and thousands of securities that don't trade very much. And because it's entirely electronic, centrally cleared, and has an NBBO [national best bid and offer], a single market; that provides dealers the opportunity to post quotes and arbitrage across those different assets. So, there'll always be a profit opportunity, but you really need those other factors, and the same thing exists with the corporate market, right? If there are different corporate bonds of different maturities—they have the same, largely the same credit risk—I can hedge out my interest rate risk; there's an arbitrage opportunity there, if they're trading out of sync. But you really need those other factors to enable that arbitrage.

Harris: Issuers can do a lot to improve their markets. First of all, they can issue fewer bonds, and they can issue simpler bonds. Both actions would improve the quality of their markets. If they improve the quality of the markets, presumably they'll ultimately be able to issue for higher prices and lower their funding costs because the option to trade is valuable to investors even if they expect that they're going to lock up the security forever, they know that they may not be able to do so.

Novick: So the irony is, why are the issuers not doing that? And the irony here is bid-ask spreads, new-issue spreads are so tight the issuers don't feel there is a need. Now, in a different market environment, they might come to the table and say, "You know, I'd like to standardize my issuance," but this whole question about is there liquidity or is there not, they would argue there's plenty of liquidity for their new issues.

I also want to add—and at the risk of being accused of selling bond ETFs to all of you—we have this great chart in the paper which shows the growth of bond ETFs—and it's not quite dollar for dollar, but it's pretty close—mirroring the decline in bond inventories of dealers. So there is a new source of liquidity here that's not based on electronic trading; it's based on an exchange. It's not the New York Stock Exchange, it's not the bond shed, but that shift is happening. It's happening as we meet, as we talk, and it's going to happen with electronic platforms, it's going to happen with bond ETFs, it's going to happen with some new thing, and, I hate to say it, but I think it's all going to be fine.

Sokobin: You will give us something to talk about next year. With that, I am going to thank the panel, and thank you all for a great day.