Speaking of Obligations

The SEC is continuing to flog the idea of imposing obligations on high frequency traders:

Regarding market structure, Schapiro said more work needs to be done to shore up investor confidence and transparency amid advances in high-speed trading.

“We are examining trading or other obligations that might be required of today’s de facto market makers: the high-frequency traders,” said Schapiro during remarks at SEC Speaks, a conference hosted by the Practising Law Institute.

“We are asking if these firms should be subject to an appropriate regulatory structure, including with respect to their quoting and trading activities.”

As I’ve said numerous times before: ain’t nothing for free.  Obligations impose costs, and the costs may be large as what Schapiro describes would constrain HFTs to offer more liquidity during times of high risk/uncertainty; if the obligations are binding, it means that HFTs must incur very high costs to supply liquidity during these periods as some prefer to make no market at all, or only very wide markets then.  They only act in that way if it is extremely risky–costly–to make markets then.

These costs must be covered.  Since entry into the market is relatively free, market makers earn a competitive rate of return on their capital.  Higher costs in some states of the world depresses that rate of return, leading to an exit of capital and higher trading costs for liquidity consumers at other times.  Thus, imposing obligations will reduce liquidity on average, and redistribute liquidity over time.

It’s also important to remember that the withdrawal of capital will offset even in volatile times the liquidity-enhancing effects of the obligations.  Yes, those market makers who remain will put a bigger fraction of their capital at risk during those periods, but there will be less of it overall to risk.  The net effect is likely to be positive during these volatile times, but probably not as much as Schapiro and others believe.  It seems that they assume that the obligations are costless and will not affect the supply of market making capital.

It is also costly to enforce these obligations.  It is costly to specify them.  This is especially true with HFT.  Liquidity has many dimensions.  How do you specify obligations on all dimensions?  What happens if crucial dimensions are left out of the rules?  How can programmers game the rules?  I have seen no serious discussion of these issues from Schapiro et al.  Economists are often–rightly–ridiculed for assuming the existence of 100 story ladders to rescue them from burning high rises, but regulators and legislators are far, far worse in this regard.

Historically, the NYSE imposed obligations on specialists.  But it compensated the specialists by giving them privileges: these privileges inflated trading costs during periods when the constraints weren’t binding.  Enforcing the constraints wasn’t easy, and in periods of extreme stress–such as 19 October, 1987–it is hard to judge whether they really performed the way their advocates would like.

Other markets–the futures exchanges, most notably, and the OTC stock and derivatives markets–have never imposed obligations.  It’s by no means clear that they’ve suffered as a result.   If these constraints are beneficial, why wouldn’t exchanges implement them unilaterally?

For once, I would like to see a serious analysis of the costs and benefits of obligations.  Schapiro keeps touting them, hyping their benefits but never even acknowledging the costs.

A serious argument would require identification of some sort of externality.  There have been papers (Greenwald-Stein in the aftermath of the ‘87 Crash comes to mind) that have done so.  But even in the context of those models, it’s not obvious that market making obligations are the best way to respond to uncertainty shocks.  Greenwald-Stein argue that trading halts–circuit breakers–can handle the problem.  In the old NYSE, specialists could call trading halts when order imbalances became extreme.  This has some advantages over price-based circuit breakers, but in the current decentralized market structure, it’s hard to observe aggregate order imbalances–ironically, this is a consequence of an earlier SEC initiative, RegNMS.

Schapiro talks a lot about market maker obligations.  Well, what about regulator obligations?  Like the obligation to make a serious case based on a well-thought out diagnosis of the problem, and a careful identification of the sources of costs and benefits, and quantification thereof.  Liquidity was in short supply during the Flash Crash, but regulators acting in such a responsible way are in chronic short supply, especially in the world of Frank-n-Dodd.