Streetwise Professor

December 17, 2011

Perfection is Not an Option

Filed under: Commodities,Derivatives,Economics,Exchanges,Politics,Regulation — The Professor @ 4:54 pm

Donkey years ago, when major exchanges were beginning the process of transforming from not-for-profit member owned firms to for-profit investor owned ones, there was a considerable amount of discussion over whether it was appropriate for for-profit exchanges to exercise regulatory functions.  The concern was that they would be subject to conflicts of interest that would undermine their incentive to regulate fraud and manipulation on their marketplaces.

I wrote a paper on the subject at the time, but shelved it in large part because these fears dissipated rather rapidly.  In the aftermath of the MFer debacle, however, they have returned with a vengeance.  Barney Frank just opined that the CME should not be a self-regulatory organization.  Yeah, I know, it’s Barney Frank (and aren’t you leaving Barney, so why not shut up?), but he’s not alone in his call.  Indeed, I get the sense that this is the emerging conventional wisdom.

So perhaps it’s time to revisit and revise that paper–thankfully Word is backward compatible!

I’ll use this post to summarize some of the arguments in that paper, and apply them to the MF/CME situation.

The main conclusion of the paper is that in comparing alternatives–self-regulating non-profit, member owned exchanges and self-regulating investor owned ones, the latter probably have stronger incentives to regulate most forms of bad conduct.

I divided the kinds of misconduct potentially subject to exchange oversight into two categories: misbehavior by agents (such as brokers cheating their customers), and market manipulation.  Based on my extensive analysis of self-regulation of manipulation, I concluded that neither type of exchange can be expected to self-regulate manipulation well.

With respect to agency problems, I concluded that for-profit exchanges are actually likely have a stronger incentive.  Member owned exchanges have more of a potential conflict of interest because the agents are among the exchange owners: some owners benefit directly from the (bad) conduct at issue.  Moreover, there is a collective action problem: perhaps one broker’s misconduct harms the others, by damaging the reputation of the exchange, but this is a public bad from the perspective of other broker-owners and free rider problems undermine the incentive of the brokers to act against it.

If you want a great example of that, look at the conduct on the floors in Chicago that was uncovered in the 1989 FBI sting.

With a for-profit exchange, however, inefficiently lax oversight of agents reduces the derived demand for the exchanges services, reducing its profits.  Put differently, for profit businesses are harmed when the suppliers of complementary services engage in misconduct that harms customers: brokers and exchanges provide complementary services, and misconduct by the former harms customers and reduces the demand for the latter’s services and hence its profits.  This gives the latter–the exchanges–a high powered incentive to police the supplier of (complementary) brokerage services.  Making them self-regulatory organizations permits them to engage in such policing and act on this incentive.

Against this it is sometimes argued that exchanges are likely to go easy on big brokerage firms that are their big customers.  But the ultimate customer is vitally important here too.  It is in the exchange’s interest that these ultimate customers are served by the most efficient, highest quality agents.  Higher cost/lower quality agents translates into lower derived demand for the exchange’s services.  Customers served by a broker that poses undue risks to its clients can go to another broker if the exchange cracks down  on the malfeasor.  The crackdown doesn’t reduce the exchange’s derived demand: it increases it, even if the broker is large.  Indeed, the risk posed by a bad broker to the exchange is all the greater, the bigger that broker is.  Which all means that there are fundamental problems with viewing the brokerage firms like MF as the exchange’s customers.

My argument depends crucially on there being a mechanism by which the self-regulating exchange incurs a cost when it engages in too little oversight.  The most likely mechanism is a reputational one, in which revelation of broker misconduct not caught by the exchange leads demanders of trading services to revise downward their estimate of the quality of the product they are getting (i.e., they conclude that they are at greater risk of being cheated by brokers), trade less (or are only willing to pay lower prices for exchange services), and thereby reduce exchange profits.

In the case of the CME and MF, this mechanism is clearly at work.  In the price of CME stock has fallen about $21/share since immediately prior to the MF blowup.  This represents a decline of market capitalization of $1.4 billion–approximately the same magnitude as the the missing money.  The work in the 1980s of Peltzman & Jarrell shows that it is quite common for product quality problems (that result in recalls) causing losses in market cap to the firm whose products are recalled that are far larger than the cost of the recall itself.  Bad quality outcomes have costly reputational effects.  This provides an incentive to avoid bad quality outcomes.

My paper drew from the property rights literature, which showed how the power of incentives could affect the incentives to supply quality.  That literature shows that a high power incentive system (like for-profit organization) could lead to a less efficient supply of quality than a low power incentive system (like non-profit organization) when quality was non-contractible or non-observable, and other mechanisms (like reputation) don’t cause a for-profit entity to internalize the effects of providing inefficiently low quality.  I concluded that this non-contractibility/non-observability issue was not relevant for the agency problems and other problems (such as informed trading).  Thus, I concluded that for-profit exchanges were likely to be better self-regulators of agent misconduct than non-profit ones.

The MF case doesn’t change my opinion on that.

But that is not necessarily the alternative being advanced by Frank et al. They presumably want to transfer all regulatory authority to the government.  To agencies like the CFTC and the SEC–which also had regulatory oversight responsibilities over MF, and didn’t stop the blowup and the possibly illegal appropriation or transfer of customer money.  Agencies that are subject to very low power incentives (except maybe for ass covering).  The CME is being pilloried for failing to stop the transfer of money that apparently began immediately after its auditors left MF: the SEC didn’t catch Allen Stanford or Bernie Madoff from stealing huge sums over periods of years–years–despite numerous warnings.

So I’m sure that yeah, handing over all regulatory responsibility to the CFTC and SEC will ensure nothing like MF ever happens again. And I also believe, as Christmas approaches, that Santa Claus is going to bring me a new Ferrari.

Actually, to extend Samuel Johnson’s aphorism, such recommendations are like 5th marriages: a triumph of idiocy over experience.

The facts in the MF case are still murky.  The collapse of a brokerage firm is not all that exceptional.  The potential theft of a billion plus in customer seg funds is very exceptional.  The issue is whether the CME could have stopped this theft.  Quite clearly such an action cannot be stopped by any ordinary and routine auditing.  Indeed, the money disappeared immediately after the last audit.  So preventing the theft would have required some extraordinary action, such as taking over control MF’s customer funds, transferring accounts, etc., earlier.  Given the truly exceptional nature of what MF apparently did, any assertion that CME should have done that reeks of hindsight bias: a common theme of the post-event commentary by industry people is that MF did the unthinkable, so it is pretty unrealistic to expect anyone to take actions to prevent the impossible (or unthinkably unlikely).

Self-regulation is imperfect.  Government regulation is imperfect.  Under either system, stuff happens.  Choosing which of these systems is, as Churchill said of democracy, the worst except for all the others that have been tried from time to time, requires a careful comparative analysis that focuses on incentives and information.  I’ve done that type of analysis in the past, and concluded that there is no single answer.  Exchanges do a lousy job at policing manipulation, and that function is best left to ex post legal enforcement–including private enforcement via civil lawsuit.  Exchanges–especially for-profit exchanges–do better at policing bad agents like brokers.  Not a perfect job, but their information and incentives are stronger in this case than in the case of manipulation.

The MF case shows that there is a market feedback mechanism at work that imposes costs on exchanges when people and firms they regulate do bad things.  Self-regulation harnesses that feedback mechanism, and it would be foolish to discard it on a whim based on an incompletely understood single instance like MF.  That’s doubly true given that the proposed alternative–complete reliance on government regulators–has more than its share of examples of proven failure, with Madoff and Stanford again being only the most prominent examples.

Remember.  Perfection is never an option.  Getting the right institutions in place requires a careful comparative analysis of the alternatives.  I can guarantee you Barney Frank hasn’t done that.  Nor have most people who are bold in their opinions on the matter. I’ve carried out such an analysis some years ago, and don’t claim to have the answer, but I know enough to say that (a) the situation is not so cut and dry and Barney et al make it, and (b) there is a colorable case that exchange self-regulation of certain forms of conduct is like what Churchill said about democracy.

And also remember that data is not plural of anecdote, and that anecdote and analysis are not synonyms.  Right now people are making strong recommendations on the basis of a single anecdote, and doing so before it is understood in any detail.  So I will withhold judgment, and when I make it I will endeavor to do so informed by the MF example, but also by a more complete comparative analysis of the incentives and information under the alternatives.

Print Friendly, PDF & Email


  1. On mf global, a huge problem is that many customers don’t read agreements and assume financial intermediaries are looking out for them. This is a common feature of people who lost money to mf global, to auction rate securities, to CDO’s, and other things.

    Comment by vbounded — December 17, 2011 @ 7:11 pm

  2. Well said, Prof. I agree with most of it, but I would add a couple of things.

    First, the exchanges’ dirty little secret from back in the member-owned SRO/CFTC-regulated days, which provided the impetus for developing an otc alternative to floor trading. Back in the days prior to the massive volumes trading otc, there was a nightly farce performed on the floor of the exchange known as “settlement.” This was the daily ritual in which prices and price curves were settled in a manner that minimized the overnight margin locals had to post to carry positions overnight. (Most locals went home flat, but the bigger guys on the floor carried size spreads, which meant they needed to minimize margin. They also had the greatest input in the settlement process). The curves were settled by the “settlement committee” in each pit, which was made up of — you guessed it — the locals who’d have to post margin. (The fcm brokers had representation in the process, but their customers didn’t have the same vested interest (minimizing margin) in the settlements — their customers usually would post margin no questions asked; settles might be challenged if they were out-of-bounds egregious, but the artful shading here and there that minimized locals’ margin typically didn’t occasion complaints from fcm guys. Besides, as a broker (not trader), you don’t want to pick a fight with a guy Monday night you might need to take the other side of a hedge Tuesday morning … your customer wants a fill, not virtue.)

    Normally, if you traded upstairs, this was a daily irritation to be endured, until you either had to buy (or sell) a physical cargo and your customer insisted the nymex settlements were reality, or you had your come-to-jesus meeting with your boss who was looking at nymex marks relative to your position: it didn’t matter that these marks were fiction, or that no one willing to transact anywhere close to the settles: Your boss was a manager reporting to a board, and his metric was the nymex settles. There is an interesting sidebar here about the settlement-price physical-market price feedback loop: Usually, sellers (or buyers) could find a willing bid (or offer) close to the settle, so the physical markets could clear around the settles (until the ready liquidity around those prices was exhausted, which requires prices to jump to a new level to clear the market … elasticities and all that …). One could come up with a model that demonstrates that such deviations away from where the majority of trading occurred introduces a small (but sometimes large) financial intermodulation distortion into the price-formation process, which might excite someone with a different instantaneous demand function to transact (assuming inelastic instantaneous supply). This could explain why price levels exhibit no central tendency and are nonstationary … but that’s a whole ‘nother issue. Who’s to say what the real price of oil was on any given day? Ultimately, though, this process redounded to the benefit of members of the settlement committee and the folks they were protecting every night. Bear in mind, the exchanges developed and enforced these rules developed by the settlement committee and got them approved by the CFTC. This was all understandable, and an audit trail could be constructed. No real harm most days, so, no real foul most days.

    Markets being markets, this hazard was corrected. Back in the mid- to late-1980s and early ’90s, the otc markets — in which strangers enter into collateralized contracts with each other — evolved as a tradeable alternative to the exchange. The otc markets developed fixed-for-floating swaps to hedge physical risk, with the floating leg being the nymex settle. Very quickly it was apparent there was enough liquidity to buy and sell swaps referencing the merc settles and to span the event horizon (i.e., the months or years you were taking on risk). The merc settlement price pretty much dropped out of the equation, since very few otc books had flat-price risk (most were big spread books).

    Residual exposure that could not be hedged otc went to the merc, but, at less than one-tenth the size of the otc market, this almost became the error term in the otc marking process. Almost, but not quite: the merc volumes were (and still are) nontrivial, and, regardless of what you think about the settlement process, it was (and, along with ICE, still is) the most ready source of liquidity at the time new information arrives to the market (i.e., a shock). The merc’s markets also provided (and still provide) arbitrage opportunities against otc exposures, and vice versa, so that, over time it would not be unreasonable to assume the prices you saw on the merc were close to being efficient (in the zero-arb sense of the word … but it was a long messy process in getting to that point).

    This is where it really gets interesting: The development of the otc markets led to increased efficiency (in the EMH sense), and prices (in levels) pretty much became random variables. This is all good. But it also means that to survive, businesses in these markets had to operate a margin businesses — profit in efficient markets cannot come from correctly guessing where prices are going. You had to be able to 1) buy on the bid and sell on the offer; 2) intermediate the bid-offer; 3) provide a forum on which such intermediation could occur; 4) develop a indispensable conduit thru which bids and offers are transmitted and collect a toll (i.e., a voice brokerage network); or 5) provide the technology that substitutes technology for labor, and allows the bidding and offering process to occur on an automated platform.

    The conversion of the exchanges to stock-ownership institutions, and the ready consolidation by the cme of those exchanges demonstrates no. 5 prevailed. But it also meant that managers from a bygone era (e.g., the guys who made their chops in the 1970s – late-90s when you could still arb huge information asymmetries) probably didn’t get the memo. They might have figured they could come in and put on a huge position and get paid for guessing correctly. When that position was put on with 40-to-1 leverage, the firm’s capital was immediately exhausted on the first down tick, and the margin required to collateralize that risk immediately exceeded the value of collateral they could post. Then, those managers had to decide whether to yak the position, which a more highly capitalized firm could do, or raid customer funds to carry it while an orderly wind-down occurred.

    Unfortunately, the market had other thoughts. And right now, the market is not only outracing firms like MF, but it is so far ahead of the folks who are supposed to be directing traffic and keeping things orderly (i.e., the regulators) that the financial intermodulation distortion threatens to produce a chaotic result (using those terms the way they’re used in physics). It’s not apparent we have the tools (or the people) to deal with that right now.

    Oh, one more thing from Alan G:

    Comment by markets.aurelius — December 18, 2011 @ 8:11 am

  3. markets, it’s not just the regulators. Most people don’t read documents. Look at MFG’s last 10-K. They said they were ramping up prop trading; they said they were counting customer collateral as a source of internal funding; they said their prop trading had liquidity risk, including on their PIIGs trades; they showed the bps sensitivity of their PIIGs trade.

    It is no wonder that many introductory brokers had been warning customers away from MFG for over 2 years.

    from edgar:

    “Historically, our trading activities have largely been limited to effecting client transactions as a broker, involving minimal principal exposure of generally short duration. As we implement our new strategic plan, we expect to increase our principal trading activities significantly, in connection with facilitating our clients’ transactions, our market-making and our proprietary activities and investing, and as a result, our exposure to market risk and trading losses will increase…. Because of the market risk and issuer risk that we face, we may incur significant losses at any time with respect to any of our principal transactions, including those executed for client facilitation and market making, proprietary activities, or asset-liability management; and particularly in the event of severe market stress.”

    “Events that would affect the funding or financing of our proprietary transactions include systemic events in the market such as, but not limited to; deterioration of broader market conditions or a global contraction of liquidity as occurred starting in 2007, as well as events unique to MF Global such as, but not limited to, a rating down-grade. In either case both funding and liquidity restrictions can result in significant losses for the firm. For example, as of March 31, 2011, we maintained an inventory of European sovereign indebtedness, which we financed using repo-to-maturity transactions; to the extent that the value of these investments decreased due to a ratings change with respect to the issuer’s long term indebtedness, we would likely be required to furnish additional margin to our counterparty.”

    “We have multiple sources of liquidity…. In addition, we have customer collateral that is not included on our balance sheet but can be re-hypothecated by us, and non-segregated customer payables, both of which may be considered (depending, among other things, on where the collateral is located and the regulatory rules applicable to the collateral) an additional layer of liquidity. Non-segregated customer cash in some jurisdictions is also available for other client liquidity demands which helps mitigate the use of our own cash. We also rely on uncommitted lines of credit from multiple sources to fund our day-to-day execution and clearing operations.”

    “The Company also enters into certain resale and repurchase transactions that mature on the same date as the underlying collateral (“reverse repo-to-maturity” and “repo-to-maturity” transactions, respectively). These transactions are accounted for as sales and purchases and accordingly the Company de-recognizes the related assets and liabilities from the consolidated balance sheets, recognizes a gain or loss on the sale/purchase of the collateral assets, and records a forward repurchase or forward resale commitment at fair value, in accordance with the accounting standard for transfers and servicing. For these specific repurchase transactions that are accounted for as sales and are de-recognized from the consolidated balance sheets, the Company maintains the exposure to the risk of default of the issuer of the underlying collateral assets, such as U.S. government securities or European sovereign debt, consisting of Italy, Spain, Belgium, Portugal and Ireland. The forward repurchase commitment represents the fair value of this exposure and is accounted for as a derivative. The value of the derivative is subject to mark to market movements which may cause volatility in the Company’s financial results until maturity of the underlying collateral at which point these instruments will be redeemed at par. At March 31, 2011, securities purchased under agreements to resell of $1,495,682, at contract value, were de-recognized. At March 31, 2011, securities sold under agreements to repurchase of $14,520,341, at contract value, were de-recognized, of which 52.6% were collateralized with European sovereign debt. At March 31, 2010, this consisted of securities purchased under agreements to resell and securities sold under agreements to repurchase of $1,199,842 and $5,702,980, respectively, at contract value.”

    “The table below presents the notional value of the issuer risk of the transactions and investments outside the trading portfolio as of March 31, 2011 net of hedging transactions we have undertaken to mitigate issuer risk. The other sovereign obligations referenced in the two below tables are issued by a group of western European countries, consisting of Italy, Spain, Belgium, Portugal and Ireland, which have a weighted average maturity of April 2012 and a final maturity of December 2012. We also constrain the tenor of country maturities as a means to mitigate issuer default risk. For example Portugal and Ireland carry a weighted average maturity of February 2012. The entire portfolio matures by December 2012, which is prior to the expiration of the European Financial Stability Facility. From time to time, and in addition to short positions in our non-trading book, we also take short positions in our trading book to mitigate our issuer credit risk further. Corporate bonds are investment grade debt of major corporations, mostly in the U.S. and U.K. [6.3B held in non-US sov debt, with 10bps delta increasing the exposure to 8.1B]”

    Comment by vbounded — December 18, 2011 @ 3:33 pm

  4. @ vbounded: great research. thank you.

    Comment by markets.aurelius — December 18, 2011 @ 3:41 pm

  5. […] Exchanges have the best incentives to be self regulatory organizations. […]

    Pingback by Tuesday Breakfast Links | Points and Figures — December 20, 2011 @ 5:19 am

RSS feed for comments on this post. TrackBack URI

Leave a comment

Powered by WordPress