Streetwise Professor

April 19, 2014

HFT, Dark Pools, Third Markets, and the Second Best

Filed under: Economics,HFT,Regulation — The Professor @ 12:05 pm

In his Atlanta Fed paper, Stiglitz uses second best considerations in his argument against HFT. My basic response is that second best considerations cut both ways.

Put simply, second best considerations mean that if one optimality condition is violated, then it may be efficiency enhancing to violate another optimality condition: or, one “market failure” can mitigate another. A simple example would be that it might be better for a polluting industry to be monopolistic or oligopolistic instead of competitive.  The monopolist’s reduction of output offsets the incentive to produce too much that occurs when there is an externality.

In the context of HFT, my second best argument is that since informed trading can be rent seeking, things that might otherwise be inefficient, such as anticipating orders or engaging in “arms races” to enhance trading speed, can be efficiency enhancing.

This is not a new theme with me. In fact, it’s quite old. I wrote a paper in 1998 titled “Third Markets and the Second Best” that applied this argument to off-exchange trading, and the free riding off of price discovery on exchanges. I discussed this further in my 2002 JLEO paper, “Securities Market Macrostructure: Property Rights and the Efficiency of Securities Trading“.

In these papers, I showed that off exchange trading venues-third markets-that free ride off of the prices produced by exchanges and limit trading to the verifiably uninformed can be efficiency enhancing even if this exacerbates adverse selection problems on the exchange because this free riding mitigates two problems: the market power of dominant exchanges (where the market power arises from the liquidity network effect) and rent seeking informed trading (i.e., the expenditure of real resources to obtain information in order to extract profits by trading with the less-informed who buy and sell for portfolio balance or risk management reasons).

Similar arguments can be applied to dark pools today. Indeed, many dark pools (and internalization) perform a similar function to third markets back in the day: they are venues that use various means to screen out informed traders, in order to reduce execution costs for the verifiably less-informed. This loss of uninformed order flow on “lit” exchanges tends to increase adverse selection costs there, but the same competition and rent seeking informed trading second best considerations arise here, meaning that the costs of lower liquidity on exchanges may be more than offset by other benefits.

And many of the very same considerations apply to HFT. Thus, contra Stiglitz, second best considerations do not unambiguously favor the adoption of restrictions on HFT.

Indeed, the thing that is most striking about the trading of financial instruments is that there are so many potential violations of optimality conditions that the entire analysis of market structure becomes an exercise in the theory of the second best.

Which can be a problem. For as George Stigler said, “Well, there are second best considerations” is a conversation stopper. But the conversation about market structure isn’t going to stop anytime soon, so we have to grasp the nettle of the second best if that conversation is going to shed more light than heat. It is good that Stiglitz makes the second best issue explicity. If only he had applied this reasoning more consistently, and recognized that informed trading can be a deviation from optimality which can be addressed by things that seem in isolation to be inefficient.

Print Friendly

2 Comments »

  1. So in the dark pools uninformed sellers transact with uninformed buyers ? That is those who are short or long an index for example….
    How do the dark pools screen for this? Is it something as simple as letting the Vanguards and SPYs in and stopping the DE Shaws at the door?

    Comment by Surya — April 19, 2014 @ 5:03 pm

  2. @surya-yes, that is the main function of dark pools. There are many screening mechanisms. Crossing networks are one example. Since the likelihood of a cross is low, an informed trader has little confidence that he will be able to profit by trading on his information. Other dark pools evaluate the profitability of the trades of their clients, and stop trading by those who make unusually large profits.

    The ProfessorComment by The Professor — April 19, 2014 @ 6:21 pm

RSS feed for comments on this post. TrackBack URI

Leave a comment

Powered by WordPress