Streetwise Professor

August 17, 2010

How Do You Like Them Apples?

Filed under: Commodities,Derivatives,Economics,Exchanges,Politics — The Professor @ 9:35 pm

I’ve written a lot about clearing mandates, but less on the Dodd-Frank Act’s attempts to alter the way derivatives trades are executed.  In part, this reflects the incredible uncertainty about just what a “swap execution facility” must be and do.

But, it is pretty evident that there is deep suspicion about bilateral OTC market dealings, both in Congress and especially among regulators.  CFTC Chairman Gary Gensler in particular has criticized bilateral trading mechanisms.  He, and many others, want to move the vast bulk of trading activity to exchange-like systems.  The standard argument is that the opaque OTC markets work to the advantage of dealers and to the disadvantage of customers.  The greater pre- and post-trade transparency in auction-based, order driven exchange markets, the story goes, will undermine the information advantage that dealers have in OTC markets, thereby eroding their profits.

As was routine in the clearing debate, this begs crucial questions.  Most importantly, why would customers choose to trade in opaque dealer markets?  Why can’t exchanges win in competition for customers with OTC dealers, if the former offer a better deal?

There is a tremendous diversity of trading mechanisms across products, and within products.  In products where exchanges compete with dealer markets, dealer market shares vary widely.  The dealer share is virtually 100 percent in FX and FX derivatives.   It is large–on the order of 80 percent–in linear interest rate derivatives.  It is closer to 50 percent in interest rate options and equity derivatives.  In the cash equity markets, a lot of trading took place on anonymous exchanges, but a good deal once took place in non-anonymous upstairs markets.  Nowadays, the upstairs markets are less important, but dark pools serve similar functions, notably facilitating large trades.  Cash Treasuries were long dealer markets, and still are to a large degree, although electronic markets have made in-roads.

There is also time series variation.  Biais and Green have an interesting paper that documents how the municipal and corporate bond markets migrated from exchange trading (on the NYSE) to a bilateral dealer market.

The simplistic “dealers scalp their poor customers” story doesn’t explain this cross sectional and time series variation in the market shares of “wholesale” dealer markets.  There must be something else going on.

What many of these off-exchange markets have in common is that they screen out informed trading.  Dark markets and third markets utilize screening technologies, or trading mechanisms that are unattractive to informed traders (e.g., crossing networks).  Others, like the upstairs markets, block markets, and OTC derivative and bond markets (govvies, corporate, and munis) rely on reputation, and are not anonymous.  Dealers reduce their risk of dealing with counterparties with private information by knowing who they trade with.  They are “opaque” in the sense that there is little pre- or post-trade transparency.  But they are far less opaque than exchange markets–especially electronic markets–in the crucial dimension of the observability of the identity of the counterparties.

The empirical evidence shows pretty conclusively that big traders incur much smaller trading costs by trading in these venues.  Which is pretty obvious: the big traders aren’t stupid, and are going to trade where they get the best prices.  The Biais-Green piece is interesting in this regard.  The market moved from exchange to OTC when insurance companies and pension funds began to dominate bond trading.

So, the claim that customers are victimized in dealer markets is hard to square with the evidence.  Big customers often benefit by trading in markets that are opaque in some dimensions, but transparent in others.  Dealer-based “wholesale” markets favor a particular kind of customer, and those customers choose to trade there.  They aren’t victims.

I am also less than persuaded that OTC market customers are all that ignorant about values.  For instance, a customer buying or selling an OTC interest rate swap can look to the Eurodollar futures market and to the Treasury futures as a source of information on the yield curve that they can utilize to evaluate the bids and offers of OTC dealers.  It’s even easier in the energy markets, to compare a dealers bid and offer on a NYMEX lookalike product with the real McCoy trading on the screen.  Moreover, those who are trading OTC have the opportunity to trade these close substitutes in highly liquid exchange markets–but they don’t.  Presumably because it is cheaper to trade off exchange than on, even though they are allegedly trading with rapacious dealers who keep them in the dark.

And this raises an issue that has been the focus of debates about “fragmentation” between exchange and dealer markets in the past, but which has strangely disappeared from sight in the current debate.  Once upon a time, a major objection to off-exchange trading was that “free rides” off of exchange price discovery, and worsens liquidity on the exchange market.  In essence, those who can verifiably demonstrate they are not privately informed can get lower trading costs by trading off exchange, thereby leaving those who cannot demonstrate that they are uninformed more vulnerable to trading against the informed, and incurring higher trading costs as a result.  That is, off exchange venues like block markets and dealer markets “skim the cream.”

This can’t be the case in some markets.  In corporates and munis, for instance, there’s no market to free ride off of, for all intents and purposes.

Even in markets where cream skimming and free riding are possible, it is not clear that the net effect of this is welfare reducing.  If the exchange market is perfectly competitive–and this is a huge if, and a huge implicit assumption in most of the literature on fragmentation and cream skimming–then the externality associated with price discovery means that off-exchange trading is welfare reducing.  But exchange markets are not necessarily perfectly competitive.  As I’ve shown in my academic work, exchanges exercise market power, through entry limits (e.g., limits on the number of exchange members), organizing member cartels, and super-marginal cost pricing.  They can do so because the “tippiness” of anonymous markets tends to make price discovery a natural monopoly.

If exchanges exercise market power, we aren’t in a first best world anymore, and therefore what would normally be considered a market failure–like an externality–can actually improve welfare.  I show this explicitly in a couple of papers, one published in the JLEO, and in a working paper titled “Third Markets and the Second Best.”  In these models, competition from off-exchange markets reduces exchange market power, leading to lower total execution costs (and higher aggregate surplus).

But even if this happens, off-exchange trading has distributive effects.  Those who can avail themselves of the off-exchange opportunities gain, those who can’t, don’t.  But the gains of the former are bigger than the losses of the latter.  These distributive effects are the reason, though, for the intense controversy over market structure.

This means that forcing trading onto exchanges can reduce competition and increase total trading costs.  Ironically, the biggest losers in this would be those who are ostensibly being helped: those who voluntarily choose to trade in dealer markets, like the OTC derivatives markets.

Some who fret about the persistence of opaque bilateral markets attribute their persistence to the “tippiness” of financial markets.  For instance, Biais and Green argue that muni and corporate bond markets have remained OTC because it is hard to tip order flow away from established markets.

I find this argument unpersuasive for two reasons.

First, and most importantly, tipping occurs in anonymous markets, as a consequence of informed trading.  The uninformed like to cluster in a single market because that minimizes their losses to the informed.  But the whole thing about many wholesale markets is that they allow the uninformed to avoid losing to the informed through another mechanism–transparency about trader identity and type.  Markets fragment between price discovery natural monopolies (where informed and uninformed trading take place) and other markets where less price discovery occurs because these trading mechanisms screen out many of the informed, making the uninformed who trade there better off.

Second, the tipping argument may be plausibly persuasive in corporates or munis, but not in many OTC derivative markets.  For instance, there are extremely liquid markets for Eurodollar and energy futures.  Traders who choose to trade OTC could avail themselves of this liquidity–but they don’t.  Again, because it is cheaper to trade someplace else.  Put differently, tipping may be able to explain something like munis where there’s only one market (though I doubt that): it cannot explain the side-by-side survival of two big markets, like Eurodollars and interest rate swaps.

In brief, there is a diversity of trading mechanisms because there is a diversity in trader types.  Forcing trading onto a one-size-fits-all platform will inevitably have distributive effects, and will almost certainly have efficiency effects too.   These efficiency effects are likely to be adverse, because anonymous trading venues where trades with private information can buy and sell tend to have market power.  Reducing competition from non-anonymous trading venues, like OTC dealer markets enhances the market power of the anonymous markets, and their owners can profit accordingly.  Moreover, in profiting, they price their services in a way that creates deadweight market power losses.

We should never expect financial markets to approximate even remotely perfectly competitive markets.  But fragmented markets in which some traders participate in face-to-face dealer markets can be better than having everybody trade on “transparent” exchanges.   This is because usual descriptions of transparency focus on price transparency, and that’s not all that matters.  Markets that offer the most price transparency often offer the least counterparty transparency.

Gary Genlser has repeatedly used the analogy of going to the store and buying an apple at a posted price to argue that everybody should want to trade on markets with pre-trade transparency.  But there is diversity even in apple markets: wholesale and retail markets are different, and offer different degrees of transparency.  The same can be said with even greater force in financial markets.

Which is why I conclude that if Gensler and those of like mind get their way, and succeed in forcing–yes, it would be by force–more trading onto anonymous, exchange-type markets, trading costs will rise and welfare will fall.  And the biggest losers will those who are the alleged victims of the lack of price transparency in OTC markets.  They will be losers because that lower price transparency is more than offset by other advantages of trading in non-anonymous markets.

Print Friendly, PDF & Email


  1. I would argue that in pure vanilla trading, a centralized exchange with less fragmented liquidity is better-like cash equities, equity options, and straight futures-along with futures options.

    I would also say that corporate bond trading is inefficient, as is muni bond trading. Information is very hard to get-and dealers control supply.

    OTC markets are different animals altogether. Mandating clearing for them might not be the best option.

    Comment by Jeffrey Carter — August 17, 2010 @ 9:52 pm

  2. I would also add that currency markets are controlled via the prime dealer network. The prime dealer network controls credit and collateral-with the power to change price depending on the situation.

    Comment by Jeffrey Carter — August 17, 2010 @ 10:36 pm

  3. […] This post was mentioned on Twitter by Craig pirrong and Craig pirrong, Craig pirrong. Craig pirrong said: Updated my SWP blog post: ( ) […]

    Pingback by Tweets that mention Streetwise Professor » How Do You Like Them Apples? -- — August 17, 2010 @ 10:40 pm

  4. But “control” is endogenous. It is the outcome of a market/competitive process; control is not assigned randomly, as in an experiment. My working assumption is that there are fundamental economic forces that determine these outcomes. If the dealer structure is so costly to customers, why can’t alternatives compete? In most markets, inefficient firms and structures get killed by more efficient competitors. Cf., the US auto industry. Why don’t you think that this same process works in financial markets.

    As I said in the post, I don’t believe that the outcome is perfectly competitive, but shutting down market and competitive processes by dictating market structure is a recipe for disaster.

    The ProfessorComment by The Professor — August 17, 2010 @ 10:44 pm

  5. The Cash Treasuries market, which is as plain vanilla as they come, has been dominated by two Inter-dealer platforms…eSpeed, and its competitor BrokerTec. The business model for these two “like exchanges” is very interesting. They charge a fixed fee to their largest customers for UNLIMITED TRADES. So unlike the CME or some other exchange, the variable cost for an incremental trade is ZERO. Which means participants can fine tune their models to capture smaller price discrepancies.

    I agree with the Professor. Anytime the regulators are going to try and force consenting adults to interact in a specific way which is different from how they interact happily today, there are problems.

    Ultimately, my belief is that the Interdealer brokers (BGC Partners, ICap, GFI Group) are in a very interesting position. They will all become SEFs under the rule, and they already have a presence in many of these markets. Some of them have significant technology capabilities…eSpeed is owned by BGC Partners and BrokerTec is owned by ICAP. They have developed the platforms to trade FX Options, CDS, and many other asset classes on a fully electronic basis. Since they have the pre-existing volume in these asset classes, I believe they will KEEP that volume when it migrates to an electronic platform.

    Comment by anthony c — August 18, 2010 @ 4:33 pm

RSS feed for comments on this post. TrackBack URI

Leave a comment

Powered by WordPress