Streetwise Professor

Articles about the energy futures and grain and oilseed futures markets make it abundantly clear that the ultimate dominance of electronic trading is drawing nigh. The commodity markets have been the last bastions of open outcry, but soon the largest of them–at NYMEX and the CBT–will begin side-by-side electronic and floor trading. If the experience in other markets is any guide, volume will move inexonerably to the electronic platform.

Although one can sympathize with the NYMEX floor traders quoted in one of the articles, realistically speaking they are whistling past the graveyard. Those leasing seats have no ownership in the exchange, and will thus have no vote and effectively no voice in the exchange’s future direction. They have no legal claim to any proceeds from the IPO. They are no different than renters of real estate with no control over the sale or conversion of their apartments.

And one piece of free advice–don’t strike, as the article suggests that some NYMEX traders are threatening. Learn from the French–as shocking as that sounds coming from me. The floor traders at MATIF (the French futures exchange that is now part of Euronext) tried that when that exchange introduced side-by-side trading in 1998. All this did was accelerate the movement towards electronic trading and led to the closure of the floor in weeks, rather than months. The striking traders exchanged a shrinking fraction of something for 100 percent of nothing. Not a good idea. The floor will continue for awhile even after the screens go live during regular trading hours. For most of the seat renters, it’s better to hang around and get some business than strike and get nothing. Moreover, seat lease prices will decline as the electronic system gains market share. Those that stick around will make a fair return on their capital. Those that strike won’t.

Although electronic trading currently has an aura of inevitability–and indeed obviousness–that wasn’t always the case. I’ve been around since the conventional wisdom in the futures markets–and the securities markets too–was that open outcry is inherently more efficient and liquid than electronic trading. Indeed, I played a role in challenging and changing that conventional wisdom. In so doing, I was subjected to scorn from some important people in the business, but as the the expression alluded to in the title says–he who laughs last laughs best.

I have followed developments in electronic trading since Globex’s development was first announced in 1986 (I was working at an FCM in Chicago at the time). Although Globex has become the orb-straddling collosus that its name connotes, it wasn’t always that way. I followed the system’s checkered early years, which were beset by disputes between the (then) partners in the endeavor, CME, CBT, and Reuters, and the opposition of many members of CME and CBT who viewed electronic trading as a threat. I remember when MATIF did the most business on Globex.

I wrote my first working paper on electronic trading in 1991 or 1992. The paper analyzed the potential efficiencies of electronic trading, emphasizing the ability of computerized markets to reduce the costs of access (thereby likely leading to higher volume) and to reduce error costs. The early paper also compared liquidity supply on electronic and open outcry exchanges. I didn’t reach any definitive conclusions on liquidity because there appeared to be potentially offsetting impacts. This was essentially an empirical issue.

I had the opportunity to begin an empirical examination of liquidity on electronic and open outcry exchanges in 1994. I was retained by Deutsche Terminborse to determine whether it was advisable for the exchange to create a new class of members analogous to locals on open outcry exchanges. DTB had heard ad nauseum that the open outcry LIFFE was inherently more liquid because of its individual market makers–locals. To determine whether DTB really needed electronic locals to allow it to compete more effectively with its London rival, I decided on a novel approach–to look at the data. The data showed that if anything, DTB was more liquid and deep than LIFFE.

Although DTB had originally intended to keep the report private and use it for internal decision making purposes only, upon seeing the results (subsequently published in a peer reviewed journal) the exchange released them publicly. The reaction was swift, and decidedly negative from some quarters–notably LIFFE. LIFFE’s CEO, Daniel Hodson, insinuated that a pointy headed academic shouldn’t be taken seriously–everybody KNOWS that open outcry is more liquid. (I’m looking through my clip files to get the exact quote from an old FT article–will update when I find it.) Other industry people were equally critical.

It therefore took some restraint on my part to avoid gloating when less than 4 years later DTB routed LIFFE in the Bund market, and Hodson was forced out as CEO. DTB used my study as part of its marketing blitz in 1997-1998. It’s fair to say, however, that LIFFE’s near death experience was more attributable to its own complacency than any academic study. When making its concerted attack on LIFFE, DTB cut prices aggressively and the British exchange did not respond until it was too late. In less than a year the Bund futures market–the largest in Europe and one of the largest in the world–had tipped from an open outcry exchange to an electronic one. No one would ever sneer at computerized markets again, and from early 1998 it was abundantly clear that the future was electronic and that open outcry markets were fighting a rear guard action.

Another working paper of mine presents a theoretical framework that helps illuminate why electronic trading has prevailed in the end. Open outcry advocates focused on two things that were right before their eyes–the pool of liquidity provided by locals on the floor, and the evident information advantage of those on the floor. Although these are features of open outcry (though the information advantage of floor traders may be exaggerated), in emphasizing these aspects, the floor partisans overlooked another important fact–the huge latent liquidity pool upstairs. The time and space advantages of those on the floor make it difficult for upstairs traders to supply liquidity in competition with the locals, especially in very volatile markets. The time it takes to send and revise orders from upstairs makes those trying to trade via limit order from off the floor vulnerable to being picked off. This taxes upstairs liquidity supply.

Electronic markets annihilate the floor’s time and space advantage–they eliminate the tax on supplying liquidity from upstairs. The dramatic reduction in the cost of accessing electronic markets has unleashed a flow of liquidity supply from what might term the “globals”–as opposed to the locals. The liquidity on the floor was visible–one could see the locals. In the floor era, the potential liquidity supply from off the floor was effectively invisible–since it was largely untapped, it was hard to understand how big it could become if the costs upstairs traders incurred to access the market were to fall dramatically. My 1994 DTB report hinted at the potential for alternative liquidity supplies, but in all honesty I too am surprised at the torrent of liquidity that computerized trading has loosed. In some respects, therefore, the skepticism about electronic trading that was so common in the 1990s was due to a failure of imagination–a failure to imagine how technological change would alter fundamentally the economics of liquidity supply.

In the old days, those on the floor were able to profit from their time, space and information advantage. Seat prices capitalized value of these advantages to the marginal exchange member–inframarginal members (the better traders, the more risk tolerant, and better capitalized ones) earned even more. The rents accruing to floor traders have disappeared as computerized markets take over, but the skyrocketing valuations of exchanges with publicly traded stock indicates that the pool of potential rents is even bigger in an electronic environment. Computerized trading, with its lower access costs, has driven substantial increases in volume. Seemingly small (but supercompetitive) fees on each trade generate large revenues for exchanges. Again, this was unimaginable only a few years ago, but now seems almost self-evident. It obviously wasn’t though. Even though I was an early supporter of electronic trading, I didn’t buy CME at the IPO or shortly thereafter. If I had only put my money where my mouth was;-)

Thinking about credit derivatives makes my head hurt–and I get the feeling that goes double or triple for most central bankers, who view them as the nitroglycerine of modern finance, powerful, useful, and highly unstable (and destabilizing). I am less concerned than they about the adverse consequences of the CD market, although the well-documented shoddy back office practices are worrisome. I have not made a detailed study of the implications of CDs for the stability of the financial system, but they are facing one problem that I do know a lot about.

Specifically, some of the recent credit blowups (e.g., Delphi, Calpine) created a situation where the supply of bonds deliverable against credit derivatives contracts was far smaller than the open interest in these contracts. This has caused disruptions in the market, and raised the specter of squeezes–and squeezes are my meat.

Here’s the basic issue. Many credit derivatives call for the bonds of a bankrupt firm to be delivered to settle the contract. For instance, if I sell default protection against company X’s bonds, if X goes belly up I have to settle my credit derivative position by delivering X’s bonds. Well, what happens when there are $10bn of CDs outstanding, but there are only $1 bn of bonds available to be delivered against them? In the case of Delphi, things were even more extreme–the open interest in Delphi CDs totaled $25 billion, but there were only $2 billion of Delphi bonds available for delivery against the CDs in the event of a default. (There is some uncertainty over what goes into this $25 billion. It may be all in single name Delphi credit default swaps. It may include, however, Delphi notionals included in collateralized default obligations and default index swaps, both of which are written on multiple names.)

If the market is competitive, that isn’t an issue. It happens in futures markets all the time. It is not unusual for the maximum open interest in a futures contract to exceed the deliverable supply by orders of magnitude. A competitive futures market can liquidate quite nicely under those conditions as long as there isn’t too much concentration in long positions. The futures contract will settle at the marginal cost of delivery/marginal value of the deliverable asset with few deliveries being made. (The lack of a centralized market mechanism for trading credit derivatives may make the liquidation process somewhat more cumbersome, but the main parties all have each others’ phone numbers, so liquidation of a large open interest under competitive conditions still shouldn’t be that challenging.)

The problem arises when one firm or colluding group of firms obtains a long position that is bigger than the supply that can be delivered at the competitive price. Under those conditions, the long can squeeze the shorts, and refuse to liquidate unless the shorts pay a supracompetitive price to settle their positions.

This could happen in the CD market. Somebody goes way long in a CD and demands delivery of more bonds than are available at the competitive price. Or goes “Texas”–i.e., goes way long the CD and the deliverables–and demands delivery of more of the remaining float. Could happen. May have already happened–It’s an issue that’s worth looking into.

Squeezes like this are an exercise of market power. In several articles and my manipulation book, I refer to them as “market power manipulations.” They might actually be more of a concern in CD markets as compared to futures markets because in the latter, when delivery will occur is known and hence market participants can mitigate (though not eliminate) market power by preparing to make delivery in advance. However, in CD markets, the timing of delivery is random–it depends on when the credit event occurs, and credit events arrive randomly. Thus, there may be a possibility of a mad scramble if a credit event occurs earlier than market participants had expected.

Many major CD players think they have come up with a solution to this problem–have an auction for the deliverable bonds, and use the auction price to settle the contract. That is, make CDs cash settled, rather than delivery settled. Working through ISDA, parties to CDs tied to Delphi that should have been settled by physical delivery agreed to implement a cash settlement mechanism instead. The cash settlement prices were based on auctions of the bonds underlying the CDs.

The conventional wisdom in derivatives markets (and not just CD markets) is that cash settlement can eliminate the possibility of a squeeze. Sorry to break it to you folks, but as usual the CW is wrong. A paper I published in the Journal of Business in 2001 shows that I can always design a delivery settled contract that is less susceptible (or at least no more susceptible) to a squeeze by large longs than any cash settled contract you design.

The intuition is pretty straightforward. A corner in a delivery settled contract works by the long demanding too many deliveries. This forces shorts to bid up the price of the deliverable above the competitive level in order to satisfy the long’s excessive demands. Similarly, somebody long a cash settled contract can force up the spot price against which the contract is settled by buying too much of the reference instrument(s) in the cash market himself. Either way, the long can liquidate the futures position at a higher price.

If there is one instrument deliverable against the delivery settled contract, and the price of the same instrument is used to settle the cash settled contract, the contracts are equally manipulable. However, if there are multiple reference issues used to determine the price of the cash settled instrument, I can design a delivery settled contract allowing delivery of these same reference issues that is less subject to manipulation.

Why? The delivery settled contract usually gives the shorts the choice of what to deliver. They will deliver the instruments whose prices are least responsive to additional buying pressure. In contrast, somebody long a cash settled contract can concentrate their buying activity on the instruments whose prices are most responsive to buying pressure in order to maximize the price impact. Put differently, when there are multiple deliverables/reference issues, cash settlement gives more power to the longs, but delivery settlement gives more power to the shorts. Since it is long market power that is the danger in most circumstances, it’s pretty obvious that delivery settlement is preferable.

As I demonstrate in my JOB piece, to be less susceptible to manipulation, the delivery settled contract must have price differentials (delivery premia and discounts) that reflect competitive price differentials. For instance, if value of deliverable A is $.10 greater in a competitive market than deliverable B, the delivery settled contract should give a $.10 premium to delivery of A. Otherwise, B is the “cheapest to deliver” and all of the large long’s delivery pressure is concentrated on that deliverable until its price is distorted by $.10. Nonetheless, a judicious choice of delivery premia/discounts results in a delivery settled contract that is less susceptible to a squeeze than a cash settled one. Put differently, the supply of bonds is more elastic in a delivery settled contract with appropriately chosen delivery differentials than is the case with a cash settled contract.

Thus, the CD fix isn’t a panacea. Indeed, it could make things worse! The next time somebody goes way long CDs on a name that defaults, the fact that the issue is cash settled will not reduce the susceptibility to the squeeze. The large long can buy a lot at the auction, drive up the price, and enhance the value of the long position. So much the better (for him) if he’s already gone Texas and can withhold securities from the auction.

Now I understand that setting delivery differentials on a CD may present some challenges. The difficulty of setting delivery differentials on government bond futures contracts has created the potential for squeezes in those markets. (Merrick, Naik & Yadav’s RFS paper “Anatomy of a Squeeze” illustrates how this happened in the UK Gilt futures market.) Nonetheless, this analysis suggests that it would be better to devote resources to specifying appropriate delivery differentials than to creating a cash settlement mechanism.

There is some information that is consistent with the hypothesis that Delphi CDs were squeezed in spite of (or because of!) the auction. According to a Nomura Fixed Income Research report, immediately after the auction the price of Delphi bonds fell from 68 to 64. This is consistent with a “burying the corpse” effect that is characteristic of a squeeze.

If squeezes like this are perceived to be a problem, deterring them will require doing something else other than cash settlement. I don’t know what legal options are available here. Would the Commodity Exchange Act’s or Securities Act’s anti-manipulation provisions apply to credit derivatives? Are CDs “commodities” under the terms of the CEA? Does the fact that a manipulation in the CD market would induce “artificial” movements in the prices of registered securities bring such conduct under the aegis of the Securities Acts? Would large market participants have recourse to lawsuits–individual or class action? Any legal eagles have any answers?

All the coverage of Alan Greenspan’s recent retirement as Chairman of the Federal Reserve Board reminded me of a story related to me by Leo Melamed, the Chairman Emeritis of the Chicago Mercantile Exchange and one of the giants of the world’s financial markets in the last half of the 20th century.

The story took place early in the morning of October 20, 1987–the day after the ’87 Crash. Along with the leaders of the other major exchanges (including the NYSE and the CBT), Melamed was a party to a phone conversation with Reagan administration officials including Secretary of the Treasury James Baker, Chairman of the President’s Council of Economic Advisors Beryl Sprinkel, and newly appointed Fed Chairman Greenspan. Baker wanted to shut the markets on the 20th in order to prevent a further meltdown. (Perhaps this isn’t surprising–Baker was never a strong free market advocate and comes off as something of a control freak.) Sprinkel would have none of it. When Baker asked for Sprinkel’s opinion, Beryl said: “We’ll close the markets when monkeys fly out of my ass!” Baker, somewhat taken aback, turned to Greenspan for support: “Well, what do you think, Alan.” To which Greenspan replied: “I go with the monkeys.”

The upshot was that the markets opened on the 20th. After a scary morning, there was a major rally in the afternoon. Some ascribe the rally to a mysterious Fed intervention. Regardless of the cause, the rally would never have occurred if Greenspan hadn’t gone with the monkeys–and if Beryl Sprinkel hadn’t had the good sense to smack down James Baker.

Zelig-like, I had some involvement in these events. At the time, I worked for a futures commission merchant, GNP Commodities. On the Friday before the Crash, I had told the company’s founder, the late Brian Monieson, that I was leaving the company because of a dispute with my direct boss. (Now you know the REAL reason for the Crash;-)} ). Brian had tried to talk me out of leaving over the weekend, but those discussions were obviously put on hold in the turmoil of Black Monday. I went into the office early Tuesday the 20th. Upon arriving at the Merc, I went directly to Brian’s office. When I saw him, he looked drawn and haggard. I said, “Brian, you look like hell. Did you get any sleep?” Brian said that he had not, and related the harrowing events of the night before.

A large firm owed the CME clearinghouse a huge sum, which if not received would have resulted in the failure of the clearinghouse. Beryl Sprinkel’s monkeys notwithstanding, such an event would have resulted in the closure of the Merc and undoubtedly the NYSE too–with consequences for the world financial system that are too frightening to contemplate.

Brian and other members of the CME braintrust spent the night trying to make sure that the payment would be made. They were assured by the firm that owed the money (Kidder Peabody, if I recall correctly) that they would make the payment. But they worried whether there would be sufficient funds to make the payment. There was little doubt that the firm was solvent, but there was concern whether they had adequate cash. As Brian put it–“They said the check was in the mail, but we wondered how high it would bounce.” That’s where Greenspan came in. He assured the Merc that the Fed would supply sufficient liquidity to ensure that the firm would have access to enough ready funds to make the payment. That was Greenspan’s first–and arguably greatest–major contribution as Fed Chairman.

Update. My memory failed me–Kidder Peabody wasn’t the firm that owed the big payment. According to Bob Tamarkin’s The Merc Kidder was actually owed money. Tamarkin doesn’t name the firm that owed the clearinghouse.