Streetwise Professor

I’ve written frequently about the problematic evolution of financial market structure under a clearing mandate.  One concern is the proliferation of CCPs.  One potential reason for this proliferation is that exchanges will want to control their own CCPs.

Long before before the crisis, I wrote about this vertical integration issue, in detail.  I dismissed worries that vertical integration is anti-competitive.  But I did note a fundamental tension.  Large, multi-product (and hence likely multi-exchange) CCPs have advantages from a risk-bearing perspective (although perhaps not from a systemic risk perspective).  But vertical integration likely economizes on transactions costs.  Thus, “siloed” CCPs can make sense even if it does not optimize risk bearing.  But the trade-off is a real one, and hard to evaluate.

I know some, notably the Bank of England, are worried about silos because they are concerned that they inflate the costs of managing counterparty risk.  But transactions costs are real costs too.

Apropos all this, today the FT has an article about how LSE is putting together its own CCP. It has cleared through LCH, but for strategic reasons thinks it necessary to control its clearing.  Given that LSE was once held out as the exemplar of how exchanges didn’t need to control their clearing, this is big news.  It also puts an exclamation point to my arguments that the evolution of the clearing business is not going to be driven purely to reduce systemic risks.  Clearing decisions are going to be made by for profit firms with their own particular agendas.  This means that the evolution of the market is likely to be quite complicated, and lead to results that the clearing mandate evangelists have not contemplated.  The implications of this structure for systemic risk are by no means obvious.

And a story in WSJ Europe raises yellow flags, and perhaps red ones, about the EU’s clearing mandate plans.  Read it. It will all sound very familiar to regular SWP readers.  Like in the US, however, it is unfortunate that people are coming to grips with the complexities of clearing mandates–and the potential dangers–only after they are a fait accompli.

Tyler Cowen wrote a post about my Cato clearing piece, for which I am grateful.  Tyler was tepid in his appraisal: and I am tepid of my appraisal of his appraisal.

On a minor point, he says I say “not necessarily” too often.  Well, I am primarily responding to arguments that say or assume things to be necessary when they are not.  Moreover, even when things are not necessarily true, they may be true, and it would be overstatement to say otherwise.  I tried to be precise in my evaluations, and neither overstate nor understate the case.

More substantively, Tyler makes the two arguments that are least likely to persuade me: arguendo ad AIG, and “look, the CME did just fine during the crisis.”  Regarding the former, I think that (a) Tyler is guilty of hindsight bias, assuming that CCPs would have seen AIG’s positions to be exceptionally risky when nobody else did; (b) this is a red herring, as AIG positions would never have been cleared; and (c) the AIG case is far more complicated, as I argued extensively on SWP many moons ago.  With regard to the latter: (a) things at CME were dicier that he lets on, something else I’ve written about, and (b) this argument is fraught with selection bias, as CME cleared the stuff that was easiest to be cleared.  Even Tyler admits that clearing CDS with their jump to default risk would be very difficult; this is true of other instruments currently traded OTC too.

My biggest objection is that Tyler really doesn’t come to grips with any of the substantive arguments in the paper, including inter alia those relating to: (a) the question of how a putatively and seriously inefficient market structure could not just survive  but thrive and grow relative to the putatively more efficient alternative; (b) the information regarding default risk and the incentives of CCPs vis a vis dealers; (c) relatedly, moral hazard and adverse selection issues; (d) the differences in costs that hedgers will bear due to cash management costs, etc., (e) the potential loss of netting economies in clearing due to fragmentation; (f) relatedly, the possibility that the industry will evolve inefficiently due to the effects of strong scope and scale economies, jurisdictional issues, and the use of clearing for strategic purposes by exchanges; (g) the ability of affected firms to alter their financial contracting (capital structures) to undo most if not all of the leverage-reducing that is purportedly one of the benefits of CCPs and more restrictive collateralization; and (h) most importantly, the potential for systemic risk arising from CCPs.  I could go on, but these are the biggies.

Tyler does, in a way, address (h), but unpersuasively, IMO.  For instance, he says: “it seems to me that clearinghouses would have had a strong profit incentive to monitor or limit the kind of risk-taking which led to the A.I.G. debacle.”  Even overlooking the AIG reference, this is problematic.  There are two cases.

The first is that CCPs are mutual organizations or effectively mutual organizations, dominated by large financial intermediaries.  In that case, the CCPs would effectively inherit the incentives of these intermediaries, which Tyler apparently thinks are inadequate to induce them to police risks efficiently.  In addition, there would be all the problems associated with moral hazard, cooperative governance, and agency problems associated with multiple principals.  (Ironically, Tyler effectively concedes this point in one of his older posts that he links to where he expresses reservations about ICE Trust.)

The second case is that CCPs are for profit firms, collecting fees, potentially in competition with one another, and taking on tail risk.  Indeed, the entire business of a CCP is taking on tail risk.  But we know that pricing that risk is hard, and that moreover, shareholder incentives often lead financial institutions to underprice that risk, or to take on too much of it (which amounts to the same thing).

So Tyler, in effect, is reduced to arguing that CCPs are easier to regulate: “The key point is that a clearinghouse can more easily be forced to carry heavy capitalization and said capitalization, unlike with a current bank, cannot so easily be undone by off-balance sheet transactions or hidden leverage.  The clearinghouse has different incentives and is much easier to regulate and therefore we should put some more trust in them.”

Somehow, putting more trust in regulators is hardly calculated to build confidence.  The point that CCPs are likely to have simpler capital structures that make it harder to take on hidden leverage is a good one, but in many respects the portfolio of risks a CCP takes on is very complicated and difficult for a regulator to appraise properly.  Again, it is all tail risk, and in a mandated environment, said risk will involve more complicated, less liquid, more difficult to evaluate instruments, and crucially, will also depend on the balance sheet risks posed by the firms it clears.  These firms can take on hidden leverage, off-balance sheet transactions, etc., so regulators have to look through the CCP back to the same institutions that allegedly are the problem in the incumbent structure.

Furthermore, as I argue above, CCP incentives are far more problematic than Tyler lets on.  Moreover, CCPs, whether for-profit firms or mutuals, have the incentive to influence regulators in the same way banks do.  Putting all this together, I don’t think it is quite as easy as Tyler does to determine the right level of CCP capital or to monitor its operations.

I’ve written many times that I do not know for sure what the optimal organization of counterparty risk allocation and pricing mechanisms is.  Nobody does.  I have often emphasized that I intend to raise issues and arguments that are grounded in good economics but which have gone unraised and unanswered in the conventional paeans to clearing.  If advocates of clearing have a good case, they could meet these arguments head on, which they have not done so far.

After reading Tyler’s piece, I have to say that most of the substantive arguments I’ve raised still remain unengaged.  So skeptical I remain.  I appreciate Tyler’s taking the time to take my piece seriously, and am encouraged that he recognizes the need for a comparative approach., but would respectfully suggest that he should take a harder look at the real problems that CCPs must confront.  I think if he does that, he’ll have more concerns about clearing mandates than he has expressed heretofore.

My first substantive post on SWP, in January of 2006, was about the pricing of message traffic on electronic trading systems.  The post was motivated by a system failure on the Tokyo Stock Exchange, when the exchange’s trading system shut down under the weight of an avalanche of orders.  I discussed why exchanges typically do not price system capacity (bandwidth), and how this can lead to problems, including system failures.

The controversy swirling around HFT has me thinking about the subject again.  In particular, the furore over “quote stuffing.” Quote stuffing is like a dedicated denial of service (DDOS) attack on a website.  Somebody submits a hugh number of orders, away from the market, and then cancels them quickly.

Quote stuffing can be done with the intent of slowing down the system on the target exchange.  This can create a discrepancy between the quotes on the targeted exchange, and the quotes on other exchanges trading the same instrument, but not targeted.  This triggers other HFTs to submit orders to the exchanges, to exploit the apparent–but not real–price difference.  The stuffer knows this is likely to happen, and can submit orders to the other exchange(s) and trade at the expense of those duped by the mirage of a price difference.

A couple of comments.  First, this is a consequence of a fragmented market structure.  The SEC chose an “information and linkages” approach over a central limit order book alternative.  The resulting fragmentation creates the incentive to exploit–or create–price discrepancies or apparent price discrepancies.

Second, free bandwidth encourages these strategies.  There have been proposals to impose transactions taxes to address this problem, but that would throw out the baby with the bath.  Not all orders cause the problem.

“Do it yourself” latency is the problem.  So charge for it.  Levy fees on message traffic that slows down system operation.  Zero prices for submitting quotes, and prices that don’t vary with their impact on system performance, encourage overuse of system resources, and also encourage opportunistic consumption of this free resource in order to extract rents from others.

The hard question is what price to charge for latency.  The cost of latency depends, in part, on the cost of the actions that others take because of the false signals emitted by systems that slow down unexpectedly and unbeknownst to them.  These firms/individuals submit orders that they would not otherwise submit, and may end up transactions that they regret once they realize that they were triggered by false signals.  It is difficult to put a number on this cost.  But it isn’t zero, so the cost of latency-inducing orders shouldn’t be zero either.

It may also be worthwhile to discriminate among orders.  For instance, a blizzard of orders far away from the market doesn’t really improve liquidity, but can impair system performance as badly as a blizzard of orders near the market.  The benefit of the away from the market orders is smaller than the more competitive ones.  Thus, it makes sense to charge more for away from the market orders that induce latency than orders that are close to the market.  The further away from the market, the bigger the charge.  This would force quote stuffers to submit orders that have a greater chance of being executed–which raises the risks and potential costs of the strategy–or to eschew the strategy altogether.

The basic lesson here isn’t really news.  Bad things happen when scarce resources aren’t priced.  So price them.

The important thing to keep in mind is to price what is really scarce.  Taxing all orders and taxing them by the same amount is non-sensical, as different orders impose different costs and provide different benefits, and these costs and benefits can vary over time.  To deter quote stuffing it is sufficient to impose fees on orders that actually impair system performance (induce latency).  To reduce the possibility that “good” orders are punished, tax orders differentially: for two sets of orders that impose the same latency, charge less competitive orders higher fees than more competitive orders.

To address the concerns with HFT, regulators and exchanges should play Bob Barker, and get the prices right.  One mystery–something that puzzled me back in 2006 when I first thought about this–is why don’t the exchanges do it themselves.  Why do they treat their scarce system resources like free goods?  Is it too costly to design and implement a pricing system?  Is it because the costs of overconsumption of system resources are borne by others?  If so, what is the best way to address the collective action problem?

I would hope, though, that in addressing this problem that exchanges and regulators focus on the central economic issue, which is pricing of a resource, and attempt to design more discriminating pricing systems, rather than impose indiscriminate rules and constraints on conduct that are likely to be both over- and under-inclusive.

A brief follow up to my earlier post on HFT and SEC Chair Mary Schapiro’s suggestion of a minimum time in force for all limit orders.

My criticism of this idea focused on one aspect of limit orders: their optionality, that exposes limit order submitters to losses if price-affecting information arrives and they cannot revise their quotes accordingly.  A time in force minimum increases the option cost of a limit order, and will induce limit order traders to quote less aggressively.  This will widen spreads and reduce depth.

This paper by Fong and Liu discusses another important aspect of quote revision: non-execution risk.  A limit order trader is not guaranteed an execution.  If the market moves away from him, his order will not be executed.  Traders who want to provide competitive quotes will revise their quotes if another trader betters them on price, or on some dimension (e.g., size) that affects secondary priority.  Thus, quote revision is part of the process of competition between liquidity suppliers (although informed traders can also be limit order traders.)

The effects of a time in force minimum on this aspect of limit orders on competition is more complicated than its effect on the option feature.  Knowing that they cannot respond to the entry of a better quote for some time, traders may quote more aggressively in the first place.  (It should be noted they can match or beat an order that under/over-cuts them, but only by entering a new order and leaving the original order in force, thereby raising their risk.)  Thus, it is difficult to determine on net whether a time in force rule intensifies competition between limit order submitters.

One last remark.  On floor-based auction markets, like futures floors, it is said that bids and offers are good only as long as the trader’s breath is warm.  This is intended to limit the options that traders provide gratis to the market.  This system has exhibited tremendous survival value, suggesting that there are some efficiency benefits to allowing liquidity suppliers to reduce their exposure.  The same principles apply in electronic markets too.  That should be remembered before requiring electronic traders to breathe long and deep, as it were.

In a speech Tuesday, SEC Chairwoman Mary Schapiro suggested imposing affirmative obligations on high frequency traders:

The SEC should consider whether traders with the “best access” to markets should be obligated to buy and sell stocks to preserve liquidity, Schapiro said today in a speech at the Economic Club of New York. The agency is also examining whether stock quotations should have to stand for a minimum amount of time. Such a change would stop high-frequency traders from repeatedly placing and canceling orders in milliseconds.

“Some could argue that May 6 was an aberration — another perfect storm — and now that it has passed markets have naturally adapted leaving no need for a comprehensive review of our market structure,” Schapiro said. “I disagree.”

Schapiro hearkened back to an earlier era, when some (and I emphasize some) market makers had affirmative obligations:

“In the old manual market structure, the market participants with the best access to the markets — the specialists and the dominant exchanges — were subject to significant trading obligations,” Schapiro said. “These traditional obligations have fallen by the wayside.”

First, a little history.  The affirmative obligation to make markets–to “lean against the wind”–was not universal.  Yes, NYSE specialists had such an obligation.  But NASDAQ market makers did not.  Nor did futures or options market makers.  Moreover, it should be noted that in exchange for their affirmative obligation, specialists were the beneficiaries of many privileges; I’ll return to this point below.

Also in a historical vein, (a) affirmative obligations are not a panacea, (b) market makers frequently attempt to avoid their obligations in times of stress, and (c) evaporating liquidity in times of stress is driven by fundamental economics, not technology, meaning that HFT-oriented markets are not uniquely vulnerable to this problem: there is, in brief, no Golden Age to which we can return through the waving of the regulatory wand.

These points are best illustrated by the events of the 1987 Crash.  The Brady Report documents that specialists were not uniformly acting as price stabilizers.  On the 19th, 26 percent of specialists were trading in a way that exacerbated price trends; on the 20th, as many specialists were trading in this way as were leaning against the wind.  The Report concluded that the behavior on the 20th reflected the fact that many specialists exhausted their capital on the 19th, which limited their ability to supply liquidity on the 20th:

The limited nature of some specialists’ contributions to price stability may be have been due to the exhaustion of their purchasing power following attempts to stabilize the markets at the open on October 19.

Moreover, even though HFT was not even a gleam in the eyes of its creators in 1987, liquidity suppliers in other markets responded to the intense uncertainty by drastically reducing the amount of liquidity they supplied.  Of NASDAQ, the Brady Report said:

During the week of October 19, some market makers formally withdrew from making markets.  In addition, some market makers ceased performing their function, merely by not answering their telephones during this period.  .  .  .  There were also widespread reports that many market makers, who normally stand ready to buy and sell hundreds and sometimes thousands of shares at their quoted prices, were only willing to fulfill their minimum obligation by buying and selling 100 shares at the quoted price.  Another indication of deterioration in market making performance is the withdrawal by some market makers from the SOES [Small Order Execution System] system, thus reducing from 1,000 to 100 the number of shares they were obligated to buy or sell.

In addition, bid-offer spreads also widened during the period.

Interestingly, the Brady report noted that although futures locals had no obligation to supply liquidity, they were “as a group somewhat more aggressive than normal in taking net positions on October 19.”  But, this herculean effort also exhausted their capital, and limited their ability to make markets late in the 19th and on the 20th.  I can attest, from personal observation, that by the afternoon of the 19th, most locals were out of the pits: they had been yanked out by their clearing firms.  Many of the remaining locals were standing with their arms crossed over their badges, to reduce the possibility that somebody would write down their ID on a trade ticket, hoping to win the outtrade lottery.

So: Don’t blame the loss of liquidity during periods of stress on technological innovation.  It happened in the horse and buggy days too.  Also: don’t expect that affirmative obligations are sufficient to ensure that markets are immune to crashes.

As I wrote in the immediate aftermath of the Flash Crash in May, the no free lunch principle works here.  The imposition of affirmative obligations to supply liquidity when it is unprofitable to do so will reduce the return on market makers’ capital.  Absent some other compensation, this will lead to the exit of market making capital.   As a result of this exit, investors will pay the cost of the affirmative obligation in the form of higher spreads/greater price impact during times when the affirmative obligation constraint is not binding.  Moreover, the exit of market making capital will offset to some degree the effects of the obligation: when the constraint binds, market makers will be risking a greater fraction of their capital to make markets during stressed times than they would in the absence of the constraint, but this greater fraction will be applied to a smaller pool of capital.  It should also be noted that one of the costs of fulfilling the obligation is that market making capital will be depleted during times of stress, thereby reducing liquidity subsequently (until new capital can flow into liquidity supply).

The previous paragraph contains the phrase “absent some other compensation.”  In the old days of the specialist system, specialists had various privileges that allowed them to trade profitably–at the expense of their customers.   These privileges were rationalized, in part, as a means of compensating specialists for their affirmative obligations.  Again: no free lunch.  Investors end up paying for the obligation.

The primary effect of affirmative obligations is to change the pattern of liquidity supply over time.  Less liquidity supply in “normal” times (due to the exit of market making capital, or profits that market makers earn from privileges extended to compensate for the costs of the obligations), less liquidity in post-stress periods, and perhaps more liquidity supply in stressed times.

It is possible that this altered pattern of liquidity supply is welfare enhancing.  But Schapiro’s remarks, and much of the post-Flash Crash commentary, does not acknowledge the costs of imposing affirmative obligations.  Thus, the proper cost-benefit trade-off question has not even been posed, let alone answered.

The old Greenwald-Stein paper written in the aftermath of the ’87 Crash theorizes that there are market failures during times of market stress, and that there is insufficient liquidity supplied during these periods.  But it is not clear that affirmative obligations are the most efficient way to address that problem.  Perhaps some other strategy, such as shifting from continuous to call trading during such periods, would be more efficient, especially inasmuch as this does not distort liquidity supply in “normal” times, as the affirmative obligation would.

It must also be recognized that defining and enforcing an affirmative obligation in the current market is much more difficult than was in the case with the NYSE under the specialist system.  The “right” obligation is by no means obvious.  HFT firms will design their systems to mitigate, not to say evade, the obligation arising under any set of criteria.  The details here will be very important, and very difficult to get right.

Another thing that Schapiro mentioned in her speech was the high rate of order cancellations, which she suggested indicate that much HFT liquidity supply is chimerical.  To address these, she mooted the idea of a minimum time for which orders must be in force.

Oi.

Limit orders are  short options.  Options are costly.  Mandating a minimum time quotes must be maintained increases the cost of those options.  Market makers will respond accordingly, by increasing spreads and reducing depth.  Again, investors will pay.  No free lunch, again.

Moreover, this recommendation cuts against the concern that liquidity dries up in stressed times.  The option inherent in a quote is more costly when markets are volatile: that’s exactly why market makers reduce their commitments during volatile times.  A minimum quote time will make market makers even more reluctant to supply quotes in volatile/stressed times.  So this measure, by itself, could make liquidity strikes more likely, not less.  To offset this, by imposing affirmative obligations, would result in higher liquidity costs in normal times: the no free lunch principle, yet again. (It would also make enforcement more difficult and costly, as market makers would have a greater incentive to evade the obligations.)

Bottom line: there is no quick fix to the evaporation of liquidity during stressed times.  Evaporation is driven by fundamental economic considerations: it is more costly to make markets during volatile periods.  Rule changes will not alter that fundamental fact.  That fundamental fact holds across trading technologies.

The costs of rules intended to increase liquidity supply in volatile times will be paid for.  Period.  The only question is when and how.  A reasoned policy debate would focus on identifying and quantifying the relevant trade-offs.  And it is my obligation to affirm that hasn’t happened yet.

On Friday I attended a conference on OTC clearing sponsored by the Chicago Fed.  The conference was excellent, and organizers Ed Nosal and Robert Steigerwald of the Fed, and Darrell Duffie of Stanford, did an excellent job of assembling a stellar group of speakers.  I had the privilege of moderating (Me?  A moderator?) one of the panels.

The speakers represented all of the major constituencies; sell side/dealers, buy side, end users, exchanges, CCPs, regulators.  Perhaps not surprisingly, there was little consensus on important issues relating to governance, organization, capital requirements, collateral, the number or size of CCPs, or any of the important matters that still remain to be decided.  It is evident that the coming months and years will be fraught with uncertainty as the industry gropes towards equilibrium in the Brave New World of Dodd-Frank.

The lunchtime speaker was Citadel founder Kenneth Griffin.  Griffin gave a paean to central clearing, and to Dodd-Frank more generally.  Clearing is cheaper operationally and administratively.  It reduces risk.  It economizes on capital.  It is the cure for all that ails the financial markets.

So if it is so wonderful, what stands in the way of its adoption?  Per Griffin: a small, self-interested cabal of dealers who reap billions and billions of profits at the expense of end users, and who will lose their ill-gotten gains in a cleared world.

This narrative is a familiar one, the stock theme of the advocates of central clearing.  And as I’ve said on numerous occasions, this story begs far more questions than it answers.  Most importantly, if the efficiencies of clearing are so manifest, and the market power losses so large, why don’t basic competitive forces undermine dealer dominance?  An inefficient business, with high monopoly markups on top of that, is a juicy target for entry by more efficient suppliers.  You would expect such a business to suffer erosion in market share; quite rapid erosion, in fact.  But the reverse has happened to dealers in recent years, as OTC markets have grown absolutely and relatively.

Maybe competition doesn’t work, or works differently, in OTC derivatives markets.  If so, it would be nice to have a convincing model, backed by empirical evidence, to explain such an anomalous outcome, so different from the received understanding in economics, and from the experience in many other industries.  Needless to say, Griffin didn’t provide either model or evidence.

He didn’t help himself by quoting OCC data about concentration in US OTC markets.  This is akin to the drunk looking for his wallet under the lamppost.  The OCC data are readily available and easy to use, but they are incomplete because they only cover US banks.  More comprehensive surveys indicate that there are currently 10 or 11 major international dealer firms; there were 14 prior to the crisis.  I’d be interested to hear Mr. Griffin identify other industries with 10-14 major competitors, and argue that they are similarly uncompetitive.

Moreover, it is well known that concentration figures can be very misleading measures of competitiveness.  They do not measure the effects of potential competition.  Moreover, concentration ratios are endogenous.  As Demsetz showed long ago, industries can become concentrated and profitable as efficient producers gain market share at the expense of less efficient rivals.

It’s not just me, folks.  After Griffin’s talk, I chatted with an economist, a former colleague with a very responsible non-academic policy and research-oriented job who was unpersuaded by Griffin’s jeremiad.   This individual said: “He ignores the IO [industrial organization] questions.”  Exactly.  Why is the industry organized this way?  Why doesn’t competition erode market power and market power rents?  Why don’t putatively more efficient organizations and institutions (clearing, in Griffin’s view) displace less efficient ones?  Why do the inefficient survive?

He brought up energy, in which clearing has made strong inroads in the past 8 years, to bolster his point.  The problem is that the case of energy suggests quite the opposite.  Clearing grew in energy naturally, organically, without a push from Dodd, or Frank, or the Fed or anybody else.  If this can happen in energy, and this is efficient, why can’t it happen in other sectors of the OTC market?

Inquiring minds, open to persuasion, want to know.  Griffin didn’t deliver.  What he did deliver was a sermon, complete with saints and devils.

Moreover, the Citadel commander’s argument presumes that end users–who, as he properly notes, must be the ultimate source of any dealer market power profits–are the victims of some sort of battered spouse syndrome.  For end users are among the most vociferous opponents of clearing mandates.  FMC Treasurer Thomas Deas was quote outspoken on this score at the session immediately prior to Griffin’s speech.  Why are end users the most ardent defenders of a system that is supposedly rigged against them, and robs them blind?

Griffin also failed to analyze seriously the trade-offs of alternative ways of allocating and pricing counterparty risk.  No suggestion of any comprehension of issues relating to the information and incentives to price and manage counterparty risk.  No mention of moral hazard or adverse selection.  No discussion of how the industry will evolve and how this will affect the ability of clearing to realize its full potential.  Most importantly, no mention of the potential systemic risks in clearing–a point mentioned prominently in the opening remarks at the conference delivered by Chicago Fed President Charles Evans.

He did acknowledge objections to clearing mandates–but in an incomplete way, and in caricature.

If the substance was, to me, unpersuasive, the delivery and tone were off-putting in the extreme.  “Supercilious” hardly does it justice.

The serious people who spoke on all sides of the clearing issue during the conference proper raised many interesting arguments and issues.  They made it plain that the way forward will be complex, and that it is by no means evident that the way mapped out by Frank-n-Dodd will improve the efficiency, stability, and safety of the financial markets.  Griffin’s talk was, in contrast,unserious, one-sided, and ultimately unconstructive.  Manichean sermons, delivered in sneering and belligerent tones, are like that.

I saw the headline of this Matt Leising article, “Lehman Derivatives Records a Mess,” and reflexively thought it would be about its OTC trades (of which there were, if memory serves, over 900,000 if memory serves).

I was wrong.  It was about its exchange traded–and cleared–positions:

Barclays PLC had no idea how big Lehman Brothers Holdings Inc.’s futures-and-options trading business was when it considered taking over the defunct bank’s derivatives trades at exchanges in 2008, a Barclays executive said.

“Lehman’s books were in such a mess that I don’t think they knew where they were,” Elizabeth James, a director of Barclays’s futures business, testified today in U.S. Bankruptcy Court in Manhattan. James worked on Barclays’s purchase of Lehman’s brokerage during the 2008 financial crisis.

She said she received an e-mail from former Barclays trading executive Stephen King saying Lehman had “absolutely no idea” if it had sold $2 billion more options than it had bought, or whether it owned $4 billion more than it had sold.

. . . .

Transferring margin together with trading positions “is normal practice in the ETD business,” she told U.S. Bankruptcy Judge James Peck, referring to Lehman’s exchange-traded derivatives. “You’d be crazy if you didn’t.”

She said Lehman’s lack of records initially prevented her from performing “due diligence” to discover what Lehman’s and its customers’ positions were, where Lehman kept its bank accounts, and who its brokers were.

James said she had seen a tabulation of Lehman’s positions at the Options Clearing Corp. showing that the daily margin requirements for its trades had varied by as much as $1 billion within a few days in September 2008.

Moral of the story: moving more trading to exchanges, exchange-like systems, and clearing will not necessarily lead to greater clarity about the positions that big entities like Lehman hold.  One of the arguments for clearing has been that it will improve regulatory transparency, and help regulators figure out where the risk is and who is exposed to whom.  Even overlooking the basic fact that it is possible to have centralized trade reporting without mutualizing credit risk through clearing, the Lehman story shows that this alleged advantage has been vastly oversold.

Big complicated firms have big, complicated, and dynamic positions that are hard to keep track of, regardless of where and how they are traded.  There is a long, long, long way to go before even approaching the ideal of “mapping” risk in the financial system.  More centralized collection of information is a necessary, but by no means sufficient, condition to achieve this goal.

The Lehman story should be a cautionary tale to all those who naively believe that waving a legislative wand and casting a spell that drives trading to exchanges will magically eliminate all of the huge operational difficulties associated with measuring and managing risk.

I’ve been writing for some time that Frank-N-Dodd will open a battle over derivatives market structure.  This battle will be waged in the markets, and in the political and regulatory arenas.  In particular, clearing mandates will lead to a protracted conflict.  The nature of the clearing business, with its strong scale and scope economies and the myriad competitive implications of alternative configurations of the clearing business will make this battle particularly intense.  And the stakes will be particularly high given the importance of clearing in the new financial architecture.

In an article forthcoming in the Journal of Applied Corporate Finance, I put it this way (link to follow when it’s available):

It is well known that the effects of central clearing on the costs of bearing counterparty risk and on systemic risk depend crucially on the equilibrium configuration of the clearing industry.  For instance, netting benefits are maximized when clearing is concentrated into a small number of multi-product clearinghouses—and arguably into a single CCP.  But the failure of an immense CCP would have disastrous consequences for the stability of the financial system, and could set off a daisy chain of failures of other CCPs, with further damaging effects.  Moreover, the scale and scope of CCPs are almost certain to affect incentives, governance costs, and the effectiveness and accuracy of risk pricing.  So it is by no means obvious what is the structure of the clearing market that optimally trades off these competing considerations.

It is equally unknown what clearing market structure will evolve, and how close this structure will come to approximating the ideal structure.  In this regard, it is essential to keep in mind two facts.

First, the extensive scale and scope economies associated with clearing make it likely that the clearing industry will be highly oligopolistic, and that strategic considerations will influence decisively the way that the industry develops.   Moreover, scope economies across trade execution and clearing (Pirrong, 2010b) will also affect the strategic and efficiency forces that will shape industry structure.  Strategic considerations will almost certainly drive a wedge between what is optimal for the individual decision makers, and what is optimal for the economy.  This is particularly true inasmuch as there is no market mechanism evident that would induce CCPs to internalize the systemic externalities associated with their failure.

Relatedly, governance and organizational form will matter.  For instance, for profit and not-for-profit mutual CCPs are likely to act differently.  Which choice is preferable?  What will determine CCP choices of organizational form?  CCPs are effectively cooperatives: how will collective action problems affect their incentives  and actions?  What regulations are required to address the governance problems that might arise under each form?  These are not easy questions to answer, but those answers will have an important effect on how clearing works in practice.

Second, industry evolution will inevitably be a highly politicized process.  Dodd-Frank gives regulators enormous discretionary authority over the operation of CCPs.  Interested parties will influence the regulatory process for their private benefit.  Political tradeoffs, rather than efficiency considerations, also threaten to cause serious divergences between the structure that evolves in practice, and the one that would optimize the relevant economic tradeoffs.  The effects of politics are particularly pronounced in this context because finance is a truly international industry, and hence jurisdictional issues and competition between jurisdictions will play a decisive role in determining the industry’s ultimate configuration.  For instance, governments in major financial centers (e.g., the US, London, the EU, and individual countries in the EU, Japan, Singapore, and Hong Kong) have all expressed a strong interest in domiciling CCPs, for both economic and political reasons.  But accommodating these interests would fragment clearing, reducing netting benefits and raising serious concerns about coordination during a crisis.  Moreover, regulatory competition between jurisdictions to favor their local CCPs could compromise financial market stability.

I’ve written elsewhere–long before the crisis, in fact–that there is a fundamental tension in clearing organization.  Risk control and moral hazard considerations provide a strong rationale to limit participation in CCPs, and/or to impose substantial capital requirements on participants.  But limiting membership to CCPs can also be a way of cartelizing the industry: given the strong scale and scope economies, a CCP that has fewer than optimal members can still be large enough to make it impossible for any other competing CCP to achieve the scale economies to compete with it.

What’s the right trade off?  Barney?  Chris?  Gary?

The politicized process is underway.  Bloomberg’s Matt Leising (article not online, it appears) and the WSJ describe how the battle lines began to form at a recent CFTC-SEC public workshop.  The competition-risk trade-off is a main axis of conflict.  First Matt Leising, describing the arguments of those who favor opening up CCPs to encourage competition:

Financial firms should be allowed to become members of over-the-counter derivatives clearinghouses even if they don’t meet current minimum capital requirements, according to an industry group. The Commodity Futures Trading Commission and Securities and Exchange Commission should allow smaller firms that don’t meet clearinghouse guidelines to become members, said Jason Kastner, vice chairman of the Swaps and Derivatives Markets Association.

Kastner spoke at a roundtable discussion today on OTC derivatives clearing at CFTC headquarters in Washington. Clearinghouse operators Intercontinental Exchange Inc., CM Group Inc. and LCH.Clearnet Ltd. in the $615 trillion privately negotiated derivatives markets enforce capital requirements and trading rules intended to bolster the ability of their organizations to withstand losses. Those requirements prevent all but the world’s largest banks, including Goldman Sachs Group Inc., JPMorgan Chase & Co. and Barclays Plc, from sending trades to the clearinghouses, the SDMA told the Federal Reserve in June.

“It’s not just about capital,” Kastner said today, referring to rules such as Intercontinental’s that members of its credit-default swaps clearinghouse have a minimum net worth of $5 billion. Regulators should create and enforce rules so that smaller firms can become members so “that it’s somehow proportional.”

But then Matt notes that some argue that Dodd-Frank is about reducing systemic risk, and has nothing to do with competition:

The main purpose of the financial-reform law, named after its principal authors, Connecticut Senator Christopher Dodd and Massachusetts Representative Barney Frank, was to make the financial system stronger by increasing risk-management standards, said Jonathan Short, a senior vice president at Intercontinental.

“It all sounds great on paper to say ‘let a thousand flowers bloom,’” he said. “I’ve never heard the Dodd-Frank Act
described as something that was done to promote competition among firms.”

[It should be noted that Dodd-Frank does include provisions encouraging competition and enjoining CCPs not to violate anti-trust rules.  How to reconcile that with the injunctions to reduce systemic risk?  Who knows?  Not Dodd and Frank’s problem.  They just wave their wands and presume the broomsticks will take care of it.]

This all makes some people nervous:

Ananda Radhakrishnan, director of the CFTC’s Division of Clearing & Intermediary Oversight, said he would hesitate before telling clearinghouses to lower their capital standards for members.

“As someone in charge of clearing, I’m reluctant to tell someone ‘you should lower the level of capital,’” he said. [In charge of clearing?  Think much of yourself, Al Haig Radhakrishnan?]

The WSJ article notes that people have picked up on the tension inherent in the issue:

James Hill, a global credit-derivatives officer at Morgan Stanley, warned that if companies without proper risk management expertise were allowed into the clearinghouses, “they will become the next “too big to fail,” and we don’t want to do that.”

After J.P. Morgan Chase & Co. managing director Jeremy Barnum said regulators would have to weigh “very complicated tensions” in determining how to write the rules, Mr. Kastner shot back that “it’s not credible to say it’s complicated.”

Not credible, indeed.

So, barely weeks after the passage of Dodd-Frank, we are seeing the sausage making process begin.  Efficiency and rent seeking and competitive considerations are all going into the hopper.  Who the hell knows what will come out the other end?  The potential for a colossal mistake–or mistakes–is great.  And these mistakes will be systemic in the effect.  For as I also say in the JACF piece:

This change is fraught with potential unintended consequences.  It is predicated on a belief that market participants systematically chose the wrong ways to trade derivatives and allocate the default risks inherent in them, but there strong reasons to believe that in fact many of the institutions adopted prior to the crisis had a strong economic rationale.  This means that forcing the implementation of a particular market infrastructure via legislative fiat runs the serious risk of reducing the efficiency of the financial markets.  In particular, although it is intended to reduce systemic risk, the mandating of expanded clearing carries its own systemic risks.  Moreover, the future evolution of market structure is highly uncertain, especially given the cost characteristics of clearing and the fact that this evolution will occur in a highly regulated environment in which political—and rent seeking—influences will play an important role.  Very bad evolutionary paths are certainly possible.  In brief, the effects of Dodd-Frank on the efficiency and stability of the financial markets are highly uncertain at best, and highly unfavorable outcomes wholly unintended by the authors are quite possible, and indeed, likely.

One final thought about systemic risk.  It is worth remembering that the imposition of a single, one-size-fits-all infrastructure is inherently systemic in its effects.  Any flaw or failure in that standard structure will have systemic consequences.  A mandate that is ill-adapted to the fundamental economic conditions in the derivatives markets creates a systemic risk because (a) ill-adaption makes failure more likely, and (b) the sweeping breadth of the mandate means that the effects of such a failure will be system wide in scope.

But have no worries, Gary Gensler tells us: “U.S. regulators won’t succumb to Wall Street efforts to weaken financial-market oversight as they implement the biggest rules overhaul since the Great Depression.”

Just how is this supposed to happen?  Yes, interested parties–and not just the big banks–have an incentive to shape the rules to benefit them.  But interested parties also have the best information and the greatest knowledge.  Are Genlser and the other regulators to shut their eyes and ears to what industry participants tell them?  That would mean ignoring the good advice, as well as the bad, self-interested kind.  If not, how are they to distinguish between the good and the bad?  Mistakes are inevitable.

This information problem is inherent in regulation, especially when involving something as incredibly complex, dynamic, reflexive/responsive, and interconnected as the financial markets.  Painting this as a morality play between the good government types (the “Goo goos” in Boston Mayor James Curley’s trenchant phrase) and the malign self-interested banks is insultingly simplistic.

Congress and Obama have given regulatory agencies the authority to lead the industry, but no map to show them where to go.  The blunder potential is huge.  The cacaphony  of voices, reflecting differences in information, interest, and motivation, will make it difficult to avoid such blunders.  Last week’s meeting gives just an inkling of the conflicts to come.

But I’m sure it will all turn out well.  It always does, doesn’t it?

You don’t have to answer that.

There was a kerfuffle earlier this week when the CFTC said that Exchange of Futures for Futures transactions–EFFs–were legal, despite the CME’s claim that they were banned wash trades.  There has been some breathless commentary that this is a grave competitive threat to CME, and raises the specter of “fungibility,” meaning that clearing and execution would effectively be unbundled.  A trade could be traded on one exchange, and then cleared elsewhere.  This would, supposedly, increase the competition that the CME would face.

I rolled this rock up the hill too many times pre-crisis, when fungibility was a headline issue before it was overshadowed by the crisis and its effects.   For instance, this was a big issue in the DOJ’s infamous passive-aggressive letter in the winter of ’08.

So, let me say it yet again.  As the Chicago School of Antitrust Economics pointed out long ago, if clearing is the source of the CME’s market power (as the advocates of fungibility claim), the CME could maximize the profits it earns by charging the appropriate price for clearing services.  It would have no incentive to tie execution and clearing.  Indeed, it would have an incentive to encourage competition in execution in order to maximize the derived demand for its (monopoly) clearing services.

So, the argument that tying execution and clearing is anti-competitive is fundamentally flawed.  There is another purpose for it.

I should also note that even with fungibility, the CME competitor ELX will face daunting obstacles in challenging CME in execution.  That’s because clearing is not the only source of scale economies in providing trading services.  Due to the effects of liquidity, there are huge scale economies in execution as well.  Price impact costs are far smaller in bigger, more liquid markets.

Indeed, it is the existence of large scale economies in clearing and execution which, in my view, drive the integration/bundling of the two services.  Integrating back-to-back monopolies economizes on transactions costs.

But integration does not require bundling of the services.  So why does CME insist on doing so?  (I.E. why do you have to trade on the CME to clear on the CME?)  The most likely explanation is that it wants control over access to the clearinghouse for risk management and control reasons.  The CME wants control over whose risk it takes and whose risk it guarantees.  EFFs could call that into question.  But if that’s the case, the CME would be advised to make it straight up, rather than relying on the wash trade argument.

I called this a kerfuffle because I don’t think EFFs will catapult ELX into contention in head-to-head competition with CME.  The CME’s liquidity advantages are too great.  This will get ELX some headlines, but not much business.

Update (1145 CT, 8/19/10). Last night, after posting this, I read something from the CME that was sent to me before I’d written the post.  It makes the “clearinghouse control” argument “straight up.”  Responding to a Consultative Paper issued by Center of European Securities Regulators, CME CEO Craig Donohue wrote:

[F]ungibility (i.e., one contract could be substituted for another contract) is not necessary or desirable.  Fungibility for OTC derivatives contracts across multiple exchanges or trading venues serves only to provide linkages, or interoperability, among various clearing houses providing clearing services for the exchanges or trading venues listing such contracts.  Interoperability among clearinghouses increases systemic risk to each clearing house and the financial system as a whole.  Indeed, when one side of a matched trade is transferred, the original clearing house automatically becomes exposed to the risk of the other clearing house.  As transfers build and links between clearing houses increase, the ability to contain a single failure decreases and the risk throughout the system increases.

So, CME is making the risk control argument in a straightforward way.

This is a good response to Bill’s question in his comment.  The key issue is one-sided transfers.  Before an EFF, trader A’s counterparty is the CCP of the original exchange (CCP1).  CCP1 has an offsetting position with A’s original trading party, B say.  If A executes an EFF with an exchange using CCP2, since both CCP1 and CCP2 have to have zero net positions, and the EFF doesn’t change CCP1’s position with B, the chain of contracts is now:  A has a position with CCP1; CCP1 has an offsetting exposure with CCP2; and CCP2 has its original exposure with B.  Thus, giving A the unilateral right to EFF into CCP1 allows him to create an inter-CCP exposure without CCP1’s consent.  CCP1 may not want to be exposed to CCP2, for both risk and operational reasons.  From a systemic risk perspective, too, this creates the dread interconnections that clearing is intended (by the ignorant, admittedly) to eliminate.

Hope that helps, Bill and BwO.

As an aside, the Donohue letter does go on to say that the CME endorses “open access” to CCPs.  The distinction is that open access as the CME uses the term means that end users can choose their clearinghouse when doing a cleared OTC deal, and if they enter into offsetting trades with different dealers, direct the trade to the original CCP in order to liquidate a position.

In response to BwO’s comment.  The argument in a nutshell is that monopoly leveraging is generally unprofitable.  Assume that product M is monopolized, and product C is competitive.  Can the monopolist increase profits by requiring that anybody who buys M also buy C from it?  Except under special conditions, the answer is no.  This is because consumers may not like the monopolist’s variety of C.  They are willing to pay less for the M+C bundle provided by the monopolist than they are for a bundle of M and their preferred variety of C.  This means that bundling reduces the demand for M, reducing the price that the monopolist can charge.  The demand for his product–and his profit–is greatest when he lets customers choose their preferred variety of C, and he charges the monopoly price corresponding to this higher demand curve.  (A similar argument holds with respect to the case where the M monopolist is not the low cost producer of C).

So, except under highly unusual circumstances (which I discuss next) it is absolutely NOT the case that “tying is anti-competitive bundling that parlays one monopoly into another monopoly.”  This has been widely understood in economics and anti-trust since the 1960s.  So yes I can argue that, and it has been argued since Bork, Posner, Director and others first addressed the issue, and virtually every model on the subject implies this result except if you build in some special (and usually contrived) conditions.  Even Whinston, who made his name on these alternative models, admits that there aren’t good practical examples of the factors he identifies leading to the anti-competitive result.

There are some counterexamples in the literature.  Hart and Tirole (1990) have one.  Whinston (1990) has several.  None of these models is remotely applicable to the case of clearing.

Carlton and Waldman (2002) have a very nice model specifically designed to address the Microsoft bundling issue in a two period model.  The key assumptions in their model is that the tied products are durable goods, and that any entrant cannot produce the monopoly product in the first period, but can produce the competitive one.  They show that under some conditions–not all–anti-competitive bundling might work here.  Neither of these assumptions is relevant in the clearing/execution context: these are not durable goods, and simultaneous entry by an integrated competitor is possible.  Nor is their assumption that one good is competitive and produced subject to constant returns to scale: in trading, both clearing and execution are subject to strong scale economies.  In this working paper I modify the Carlton-Waldman model, and show that in conditions that are most plausible in the clearing/execution setting, anti-competitive bundling is not likely to be a profit increasing strategy for the clearing monopolist.  That paper discusses the literature in some detail, and examines other transactions-cost based reasons for integration of clearing and execution.

Thanks for the comments.

In a semi-follow up to yesterday’s post on trading mechanisms, today I’m going to explore post-trade price transparency.

One example that has been brought up repeatedly in the context of increasing transparency in the OTC derivatives markets is the introduction of TRACE in the corporate bond market.  TRACE mandated the disclosure of all corporate bond trades, including information on price and trade size.  Currently, the information has to be reported within 15 minutes of the trade.

It has been argued that TRACE improved the efficiency of the corporate bond market, and reduced trading costs and dealer profits.  This supposedly occurred because the opacity of the market pre-TRACE gave the dealers market power that they could exploit to their benefit.

The empirical basis for these pro-TRACE claims is a couple of studies.  One compared round-trip trading costs pre- and post-TRACE, and found that post-TRACE costs were smaller.  Another exploited the fact that TRACE was phased in over time, and ran cross-sectional regressions that compared trading costs for bonds that were TRACE-eligible and those that weren’t, while controlling for other observable factors.  This study also found that TRACE lowered costs.

I am always a little leery about these kinds of studies.  They implicitly assume that the characteristics of the transactions in the TRACE and non-TRACE samples are the same, or can be controlled for based on observable variables.  I’m not quite so sure, especially given some of the other things that transpired post-TRACE–as I’ll discuss now.  (In social sciences, you always have to worry about endogeneity problems.  They’re devilishly hard to control for; natural experiments or good instrumental variables are hard to come by.)

Most notably, there is anecdotal evidence that post-TRACE, corporate bond dealers dramatically reduced the size of their trading teams, and reduced their inventory holdings.  They effectively transitioned from dealers to brokers.  The reduction in inventory means that they reduced their capital commitments to this business.

Now this is hard to square with a story that TRACE reduced dealer market power, which in turn reduced trading costs.  Typically, firms exercise market power by restricting output and capacity.  Thus, something that increases competition should lead to an expansion of output and capacity.   The opposite happened, apparently.

You can tell a kind of monopolistic competition story.  Dealers had market power due to opacity, but competed away the rents by overinvesting in capacity, or in “quality” by investing excessively in inventory that allowed them to accommodate customer needs more readily.  But even in a monopolistic competition story, increasing competition (by reducing customer search costs, for instance, through better disclosure) makes individual firm demand curves more elastic, meaning that those firms that continue to operate should increase output and operate closer to their minimum average cost point.  That’s hard to square with what happened post-TRACE.

It should also be noted that the theoretical foundation for the contention that improved disclosure reduces profitability by increasing competition is weak.  Madhavan (RFS, 1996) has a sequential trading model in which dealers can choose not to disclose trades, but are in competition with one another.  In equilibrium, dealers don’t disclose a trade that occurs today because the order flow is informative, and the information they get when they do a trade allows them to trade more profitably in the future; a dealer that trades today has a competitive advantage over other dealers in trading in the future.  But in equilibrium, dealers don’t earn any profits.  This is because they compete aggressively to attract order flow today.  Some trades are “loss leaders” that they use to attract order flow that increases the profitability of future trades.  Thus, lack of disclosure is not necessarily profitable for dealers because competition for market power dissipates rents.

There are other post-TRACE developments that are hard to square with the Gary Gensler & apple pie story.  (The details can be found in a JEP paper by Bessembinder and Maxwell.)  Notably, if you believe that trading costs really fell, this should have benefited corporate issuers.  Ultimately, investors are willing to pay less for bonds, the greater the costs of trading them later.  If trading costs fell, publicly traded bonds should have become a relatively more attractive form of finance for corporations.  But in fact, post-TRACE, firms relied much more on privately placed bonds; corporate bond issuance fell; and CLOs (collateralized loan obligations–securities backed by bank loans) grew dramatically.  So, corporations relied much less heavily on instruments that should have become relatively cheaper to issue if trading costs really fell as a result of TRACE, and relied more on stuff that couldn’t be traded at all (private placements) or traded in dealer markets (the CLOs)!  This is inconsistent with the view that transparency reduced corporate bond trading costs dramatically, and rather implies the opposite.  (Of course, a lot of other stuff was going on during this period, so it’s a stretch to lay all this on TRACE, but still, it is facially not supportive of the claim that TRACE reduced trading costs.)  (This also raises the question: if the dealer structure without disclosure was so inefficient, why didn’t issuers use their leverage to mandate disclosure, e.g., by conditioning participation in underwriting syndicates to firms that committed to disclose trade information?)

Why might trading costs have gone up?  Anecdotally, it became harder to trade in size post-TRACE as dealers weren’t willing to take on big positions once they had to disclose that they had done so.  That is, dealers were at risk of being front run once they took on big positions from a customer/counterparty.  So they offered less liquidity, making it more expensive for the insurance companies and pension funds that dominate the market to trade.  This would explain why they reduced their head counts, cut back on inventory, and essentially shifted from dealing to broking.

This all suggests that the effect of TRACE was to lower the transactions costs for some (primarily small retail traders), increase trading costs for others (mainly big investors), and change the composition of trades (in favor of trades that were easy to broker).  Put differently, TRACE made “wholesale” trades more expensive–pricing some out of the market altogether.  Not an unalloyed benefit, and arguably quite the reverse.  If the corporate bond market is primarily a wholesale market, the effects of transparency could have been detrimental on net, even if some benefitted.)

This isn’t dispositive, but it should give pause.  In particular, when evaluating the effect of a market structure change, or of a mandated change in transparency, you need to look beyond simple before and after or cross sectional price comparisons, as these may be incomplete and misleading.  Moreover, you need to look at things other than price.  If a policy change results in exit of firms and reductions in the size of firms, and substitution away from the product affected by the change, you have to seriously question whether the change increased competition.  Usually you’d expect reductions in barriers to competition to lead to entry, increased output, and substitution towards the product that became more competitive.   When the exact opposite happens, you better have a really good story–which I haven’t seen anybody advance–or you better seriously entertain the hypothesis that the supposedly pro-competitive change had the exact opposite effect.

So, rather than being the feel good story of the summer, upon closer examination TRACE appears to be an ambiguous tale at best, and more likely a highly cautionary one.  And those who constantly prattle about the wonders of transparency–and you know whom I speaking of–should be pressed to address all of the dimensions of the TRACE experience.   Because it is quite plausible that interventions into wholesale markets, including mandates to make trading more “exchange like” with more pre- and post-trade transparency, can increase the costs that customers in those markets pay.