Streetwise Professor

April 23, 2014

File Under “Dog Bites Man”: Exchange Monopolies and Dark Pools

Filed under: Commodities,Derivatives,Economics,Exchanges,Politics,Regulation — The Professor @ 2:13 pm

An exchange chairman believes that all trading should take place on exchanges. In commenting on securities market structure, CME Group Chairman Terry Duffy criticizes fragmentation-especially the existence of dark pools-and touts the lack of fragmentation in futures trading.

The concentration of trading activity on futures exchanges, as opposed to the fragmentation across different exchanges (as well as off-exchange venues) in equities is due to a major difference in the treatment of orders. In futures markets, exchanges own their order flow: hypothetically, if there was another exchange posting a better price in a particular product, CME would not be obligated to direct an order to that better-priced market. When exchanges own their order flow in this way, traders direct orders to the exchange where they expect to get the best price. This is typically the market where most traders are. This creates a centripetal force that causes all activity to tip to a single dominant exchange. That is why CME, Eurex, ICE, etc., have monopolies or near monopolies in the products they trade. (And yes, Terry, even though no one is stopping anyone from competing with you, this order flow effect means that no one can do so effectively, leaving you a de facto monopoly. Only LIFFE’s idiocy in its battle with Eurex in 1998 allowed the Germans to get trading in the Bund futures to tip their way.)

This is the way it used to be in equities too. Prior to the late-2000s, the NYSE effectively owned its order flow, and 80-85 percent of trading volume in NYSE listings took place on the NYSE. The remainder occurred on “third markets” that catered to the verifiably uninformed (more on this below).  But in 2005 the SEC changed the rules in a fundamental way. It passed RegNMS, which socialized order flow by requiring exchanges to route orders to others displaying better prices. Within a very short period, a handful of exchanges executing between 8-20 percent of volume competed fiercely with one another. The NYSE’s effective monopoly had been broken.  This is why Goldman paid $6.5 billion for a specialist unit in 2000, and sold it for $30 million this year. The 2000 price capitalized monopoly rents: there are none to capitalize in 2014.

Duffy says he’s fine with this kind of fragmentation  of trading across exchanges with the associated intense competition (though that’s very easy for him to say because he doesn’t have to worry about that outcome given the lack of a RegNMS-type rule in futures markets), but he thinks dark pools should be shut down.

To evaluate this position, you need to understand what role dark pools play. Just like third markets and block markets of the pre-RegNMS era, dark pools (and internalization of retail order flow) are a ways of screening out informed traders. This reduces the costs of the uninformed who can trade on dark pools be reducing their vulnerability to adverse selection. This is good for them, but the overall effects are much harder to understand. Order flow on exchanges becomes more toxic (i.e., a higher proportion of the order flow is informed) which raises adverse selection costs on exchanges, and thereby raises trading costs there.

The net effect of this is very difficult to determine. This is another application of the second best. Since exchanges may have market power, the additional competition from off-exchange venues can improve efficiency even if it raises adverse selection costs for some traders. Moreover, as I’ve argued in my HFT posts recently, since some informed trading is of the rent seeking variety, by reducing the returns to informed trading dark pools can reduce wasteful investments in information.

This means that Duffy’s criticism of dark pools might be right. But it might be wrong.

One thing is definitely true. Market structure has huge distributive effects. Although the rules on dark pools have very uncertain efficiency effects, there is no doubt that these rules affect the distribution of costs and benefits across different types of traders. It is precisely these distributive effects which make the battles over market structure so divisive and protracted.

I’d also note that Duffy ignoring some features of futures markets, and derivatives markets generally, that perform functions similar to dark pools. For instance, CME allows block trading. Indeed, it is engaged in a tussle with the CFTC, which wants to reduce the amount of block trading in order to force more volume into the order book.

But block trades are a way that less-informed large traders can reduce adverse selection costs. They have long performed this function in equity markets, and are now doing so in futures. And by stripping out that order flow from the order book, block trades have the same effects as dark pools.  Blocks are a form of fragmentation.

Block markets are non-anonymous: that’s how they screen out the informed. Block traders won’t deal with those they believe likely to be informed, and by trading face-to-face traders can develop reputations for not being informed and profiting systematically at the expense of their counterparties.

Well, wouldn’t you know it, but this is how OTC derivatives markets work too. The lack of price transparency in OTC markets is often bewailed, but OTC markets are transparent in another important way that exchanges are not: they offer counterparty transparency, whereas exchanges are counterparty opaque. This benefits, say, firms that are trading to hedge in large volume (who are likely to be uninformed). It’s not a surprise that trading activity migrated from OTC to blocks on CME and ICE after Frankendodd made swaps trading more expensive. Both futures blocks and swaps are ways of reducing the execution costs of large, likely uninformed traders.

Put differently: blocks (and swaps) are a form of fragmentation, in the sense that they divert trading activity away from the limit order book. So Duffy shouldn’t be quite so sure about the superiority of the futures market model. It is fragmented in its own way, and has a lot more market power. But of course Duffy likes the last part, though he would never admit it.



Print Friendly, PDF & Email


  1. As I keep reading your posts on exchanges and HFT, suddenly it dawned upon me what the intention behind IEX – touted in “Flash Boys” might be. I think informed investors might find it appealing to trade with uninformeds like Vanguards or SPDR ETFs. It would be particularly convenient to have a pool that excludes HFTs. So, if the uninformeds are given a justifiable reason, they might be induced to join trading with the informed. And they might get taken out for dinner by the informed even! It just needs some planning to put something out there that sounds “reasonable” on paper…and Michael Lewis the ex-bond salesman is good at doing just that.

    Btw, I am awaiting your post on Piketty’s hefty tome that has just displaced Flash Boys from the top of Amazon’s list of best sellers.

    Comment by Surya — April 23, 2014 @ 8:14 pm

  2. by HFTs…I meant those that are good at sniffing out informed traders & adjust quotes quickly to deal with adverse selection. I think it goes without saying that anyone who survives & remains profitable would be good at doing this.

    Comment by Surya — April 23, 2014 @ 8:19 pm

  3. Saw your research piece for Trafi, it was featured on Reuters 🙂

    Comment by Surya — April 24, 2014 @ 9:38 am

  4. Or to put it more bluntly Craig, Mr Duffy wants all trades on his Exchange to increase his own revenues. A slight conflict of interest as you say, in similar fashion to Exchange bosses also arguing co-location does not harm the fairness of the market as paying for “speedy boarding” is open to everyone, whilst simultaneously collecting the related revenues.

    That said, I do concur that the challenge with dark pools is that they are parasitical in nature in using prices from lit markets and don’t contribute to price formation. With the average trade sizes in dark pools having fallen to small tens of thousands of dollars it is also getting much harder to suggest that they largely exist to avoid market impact on block trades.

    Lastly, one crucial aspect re Futures liquidity concentration in venues is IP ownership and clearing of the related open inteest – an equity can be traded on multiple venues because there is no “lock-up” on the product to a single Exchange. In contrast the IP on Futures is treasured by Exchanges in providing incumbents (with existing open interest) a huge barrier to entry by competitors i.e. they don’t allow competitors to trade the same product nor open up access to their open interest at the in-house CCP. Whilst you mention the success of Eurex is luring away liquidity from Liffe, there are relatively few similar examples of new look-alike products achieving the same result. Perhaps the nearest contemporary story is NLX pulling in approx 10% market share on a regular basis in London, but against the context of only recently having introducing small (trivial) fees, an uplift from zero which previously applied.

    Comment by John Wilson — April 24, 2014 @ 10:13 am

  5. He is right about market structure and HFT. Because we have shitty structure on the equity side, HFT causes more distortions. You don’t see Nanex tweeting too much about distortions in price on futures.

    Futures HFT issue is that the exchanges have created 1st, 2nd, 3rd degree price discrimination by selling co-location. Reporter should have asked them why Duffy sells an advantage to one piece of the market. Should have also asked why certain firms get rebates, and some don’t causing competitive misalignment.

    Comment by Jeffrey Carter (@pointsnfigures) — April 26, 2014 @ 9:12 am

RSS feed for comments on this post. TrackBack URI

Leave a comment

Powered by WordPress