CME CEO Craig Donahue sent on some thoughtful remarks on the evolution of financial market structure that he gave at an International Monetary Conference that merit some comment.
A crucial point he makes is how demutualization has changed the nature of the relationship between exchanges and intermediaries, notably the large FCMs. Whereas the FCMs were once exchange members, and hence owners, and exchange efforts were devoted in part to enhancing their profits, now exchanges are focused on maximizing returns to shareholders. This, as Donahue puts it, creates a tension between “value chain players” with exchanges and intermediaries both chasing dollars from ultimate customers. Although this is not a zero sum game, exchanges can profit at intermediaries’ expense, and vice versa. Exchanges and intermediaries provide complementary services, and each is striving to get the biggest piece of what customers are willing to pay to execute and clear a trade.
This is not to say that these conflicts never existed. Donahue notes that non-profit exchanges were “inefficient member-run utilities.” In large part, that inefficiency was a feature, not a bug. Inefficient, committee-dominated, deliberative decision processes were in large part a way of managing tensions and conflicts and rent seeking between different types of exchange members, such as FCMs, floor brokers, and locals, all of whom were “value chain players,” and each of whom had an incentive to seek rents from other parts of the value chain. Thus, the tensions between exchanges and intermediaries are just now more out in the open. They used to be all in the exchange member family, as it were, but we all know that intra-family disputes can sometimes be very intense indeed. (I saw some of these battles up close when I worked with the Chicago Board of Trade to redesign its grain futures contracts, first in the early 1990s, then in 1997. I saw short hedgers pitted against long hedgers, Locals pitted against brokers, etc. Committee gridlock was intended in part to ensure that nobody could do anything to drastically change the status quo, and hence reallocate profits from one group of members to another.)
Donahue also notes that in the demutualized world, exchanges are venturing into the intermediaries’ space (e.g., the CME FXMarket Space and Swapstream) and intermediaries are striving to execute trades, either through internalization or the creation of execution platforms. He pays particular attention to the issue of “dark pools,” a subject that Clara Furse has been quite outspoken on as well.
Dark pools are off-exchange execution vehicles. Importantly, they typically do NOT contribute to price discovery, instead allowing large traders to execute orders at prices derived from those established on the exchanges where price discovery does occur.
Donahue (and Furse as well) raise concerns about the impact of this fragmentation on price discovery and the liquidity of exchange markets. The basic idea–which has been understood for quite awhile–is that dark pools (once called “third markets”) siphon off uninformed traders who provide liquidity from exchanges. This exacerbates the adverse selection problem on the central exchange market, leading to wider spreads and lower depth. The dark pools free ride off of exchange price discovery, and actually impair the quality of that price discovery by reducing the amount of trading activity that occurs on exchanges where the price discovery actually takes place.
It is quite correct that fragmentation impairs liquidity on central exchanges where price discovery occurs. It may be going to far, however, to conclude that this is inefficient, at least relative to the alternative of discouraging the formation of dark pools/third markets. Third markets may be second best, as I argued in a paper with a similar title, and in an article published in JLEO in 2002.
The argument is that central exchanges are likely to possess market power due to the network effects of liquidity. The existence of exchange market power means that we are not in a first best world, and hence the theory of the second best implies that an externality (such as free riding off of exchange prices) does not necessarily reduce welfare relative to what it would be if the externality was eliminated. (If absent the externality the exchange would supply the optimal level of output, the externality would reduce surplus.)
In the formal model in these papers the dark pools/third markets actually increase total surplus because they result in increased trading activity–trading activity that would be too small in the absence of the third market due to the output-restricting policies of exchanges with market power. The defection of some traders to third markets indeed reduces liquidity on the central locus of price discovery, making the traders on the central exchange worse off, but in the model the defecting traders reduce their trading costs (price impact costs) by an amount that is larger than the increase in trading costs on the exchange.
Thus, theoretical models do not support a blanket conclusion that fragmentation and free riding on exchange price discovery is a bad thing. I acknowledge that these results are dependent on the assumptions underlying the model, and that different results might obtain with different assumptions. The models do serve to caution us, however, that fragmentation is not necessarily a bad thing if competition between exchanges is imperfect, as is almost certainly the case due to the centripetal force of liquidity that tends to make price discovery a natural monopoly.
In my view, the dizzying pace of change in market structure in derivatives and securities trading is a very complex phenomenon. Moreover, it is very difficult to draw hard and fast conclusions about the desirability of various alternative arrangements because network effects and the information-intensive nature of trading mean that the “perfect competition” benchmark of textbook economics is not achievable in financial markets. All alternatives involve trade offs, and arguments over which arrangement is second best, third best, fourth best . . . . The devil is truly in the details, and it is very important to understand the microfoundations of trading. Different formulations of those microfoundations can lead to very different conclusions. The formulation of the microfoundations in my models leads to one conclusion–other formulations might lead to opposite implications.
In such an environment, it is important to have vigorous–and thoughtful–debate so that the relevant issues and trade-offs can be identified and evaluated. Craig Donahue’s speech is a fine example of such a thoughtful contribution, and I encourage anyone interested in the evolution of financial markets to read it when it is published (as I understand it will be soon.) In the meantime, it has reinvigorated my interest in these issues, and I will devote some effort to pushing the theoretical envelope to help gain a better understanding of how fragmentation affects welfare under alternative assumptions about microfoundations. My earlier work was inspired by mutualized exchanges that exercised market power by limiting membership rather than by charging supercompetitive prices for their services. Things may be different in a demutualized world where exchanges have an incentive to encourage entry of liquidity suppliers, and exercise market power by charging higher prices rather than by limiting access. The key thing will be to come up with models of such exchanges that predict that dark pools will form in equilibrium (as is clearly the case), and then to see how trading costs and trader surplus depends on policies affecting the operations of third markets. So many interesting problems, so little time!