Streetwise Professor

Regulators seem really, really scared of the dark.  Members of Congress speak in ominous tones about “dark markets” when referring to OTC markets.  Another “dark” subject that has been the subject of horror stories in the recent past–“dark pools”–is making a reappearance.  Last week the Federation of Securities Exchanges wrote a letter to regulators demanding that dark pools be regulated more extensively, and that measures be implemented to move more trading onto exchanges.  In today’s FT, Jeremy Grant reports on regulatory pique at dark pools:

Dark pools operated by banks should improve pre and post-trade transparency and should operate under a more formal structure, the head of Europe’s umbrella group for securities regulators has said.

The comments, by Eddy Wymeersch, chairman of the Committee of European Securities Regulators, are the first by a top regulator since a spat has developed between the banks and some of Europe’s biggest exchanges over the bank’s “crossing networks”, a type of dark pool. Dark pools allow the matching of large blocks of shares without prices being revealed until after trades are completed.

Regulators on both sides of the Atlantic are studying them amid questions over their transparency.

. . . .

Mr Wymeersch said bank dark pools should “at least have post-trade transparency and also, if possible, have pre-trade transparency. They should organise themselves as MTFs, or something similar”.

“It’s a question of reliability of valuation in the entire system,” Mr Wymeersch told the FT at a European Commission conference on over-the-counter derivatives in Brussels.

He said he was “concerned” at the amount of off-exchange equities trading that takes place on crossing netwo

The criticism about dark pools relates to issues of transparency and fragmentation.  It is commonly argued that by fragmenting trading and liquidity, and in particular by siphoning off relatively uninformed order flow, dark pools degrade price discovery on exchanges.

This may be true, but it raises the question as to why it would occur, and if it is necessarily inefficient (relative to realistic alternatives).  I’ve written about this a good deal in my academic research, including a paper published in the Journal of Law Economics and Organization, another in Regulation Magazine, and a couple of working papers, and have raised some issues that most policy discussions overlook.  I’ll try to recap them briefly here.

Dark pools are essentially venues where large traders send orders in the belief that their orders would have a smaller price impact if executed in the dark pool, than if submitted to the exchange. That is, these traders believe that their orders will be less costly to execute.

That is quite possible if dark pools operate effective mechanisms for screening out relatively informed traders.  The crossing networks described above are one way of doing that.  Those submitting orders to crossing networks are not guaranteed an execution; only if there is a contra order that can be matched will the order be executed.  Moreover, crossing networks frequently cross orders periodically, rather than continuously.  These features are unattractive to an informed trader.  If I think I have very good information, I want to make sure that order is executed.  Moreover, if the information is likely to become public shortly, I don’t want to take the risk it will come out before the next cross.

This means that crossing networks can help separate uninformed block traders from informed traders.  As a result, the crossing network is less susceptible to adverse selection; this tends to reduce price impact of large orders.

In this respect, dark pools that screen out informed traders in order to reduce trading costs for large, uninformed traders (e.g., index funds) are no different than upstairs block markets that have been around for decades.  Upstairs block trading was a way of reducing execution costs by screening out the informed: if somebody “bagged the street” and traded a big block right before information was released, that somebody wouldn’t be able to trade blocks again anytime soon.

Theoretically, this “cream skimming” makes markets that don’t screen less liquid: there are fewer uninformed traders (because some have been skimmed off by the dark pools) on the exchanges that don’t screen, so adverse selection problems are more severe there.

Similar issues arise in other activities that draw regulator suspicion, such as payment for order flow and internalization.  Market makers like to internalize orders that they know are less likely to be informed, and send the others onto the exchange.  Similarly, they buy order flow that is likely to be uninformed and execute those orders.

In a world where exchanges are perfectly competitive, this cream skimming would be inefficient because it causes an negative externality imposed on those who trade on exchanges.  But that’s not likely to be the real world in which we operate.  Traditionally, exchanges limited participation (for instance, the number of members on the NYSE didn’t change from 1929 on–it remained at 1366, if memory serves, during that entire period).  Entry barriers reduce competition in market making.  Now, exchanges likely have some market power due to the network effects of liquidity, and can exploit that market power by charging supercompetitive prices.

My academic work shows that in such a world, cream skimming (e.g., the formation of dark pools) is profitable, and what’s more, it can improve efficiency in the sense that it reduces total investor trading costs.  Those that trade on exchanges are made worse off because their trading costs go up, but the execution costs of those who can take advantage of the third market that screens out the informed fall by a larger amount.

These effects occur because the competitive impact of dark pools more than offsets the negative externality.  Dark pools (and other off-exchange venues) enhance competition, reducing deadweight losses arising from market power.

Moreover, dark pools/third markets can reduce the return to informed trading.  Since some expenditures to become informed are pure rent seeking (the informed just extract a rent from the uninformed), reducing the incentive to collect information reduces rent seeking costs.

Thus, the pros and cons of dark pools or other off exchange trading venues that “fragment” trading depend crucially on the competitiveness of exchange trading.  Given that the network effects of liquidity are likely to confer market power on a dominant exchange, it is quite possible, and indeed probable in my view, that dark pools/third markets/off-exchange trading venues are a “second best” response to this market power.  If this is the case, constraining third markets/dark pools may just enhance exchange market power by disabling their competitors.  No wonder exchanges are the most outspoken in their criticisms of dark pools, and the biggest cheerleaders for more regulation.  Go figure!

Put differently, dark pools and other off exchange venues compete against exchanges, and since exchanges are likely to possess market power, this competition can improve welfare even if these dark pools create a negative (liquidity) externality.  That is, the arguments against dark pools focus on the negative externality, which would lead to inefficiency if the market were otherwise perfectly competitive.  But once deviations from perfect competition are admitted, the efficiency case against dark pools and other off-exchange venues fails.  The cost of externality may be smaller than the benefit of greater competition.

This means that any evaluation of the effect of dark pools, and the costs of benefits of alternative regulatory actions, is an inherently complicated exercise.  You can’t just say, “dark pools create a negative externality” and stop there.  You have to evaluate the trade-offs between the liquidity externalities and the competitive effects.  This is a very challenging task, and one that is likely beyond the capability of regulators to get right.

Suffice it to say that the complaints of exchange operators about dark pools should not be taken at face value.  Their gripes are self-interested.  Hobbling dark pools would reduce the competition they face.

Getting market structure right in securities markets is hard because of the network/liquidity effects.  Ham-handed regulatory interventions based on simplistic analyses of externalities and transparency are likely to make things worse, rather than better.