Last week, Professor Charles M. Jones, a noted economist at Columbia, published an opinion piece in POLITICO claiming to enlighten readers on the realities of high-frequency trading (or "HFT"), computer driven trading at millisecond speeds driven by complex algorithms based on complex trading strategies. This has surfaced as the subject of politically charged debate in the context of of a proposed financial transaction tax that would, among other things, curb the most excessive forms of HFT.
The POLITICO piece and a longer, academic-style article were published simultaneously. These articles catalogue existing claims of the benefits of HFT and dismiss concerns of the potential harm of such activity.
Perhaps the most important bit of information does not appear in the articles. It can be found at the very end of the press release by Columbia Business School announcing the article: "The research was supported by a grant from Citadel LLC." The interest of Citadel in the subject is clear. Citadel is one of the largest HFT firms, its funds returning as much as $1.1 billion a year from the business.
Citadel's largesse does not mean that Professor Jones' work was biased. But the very fact that he published an opinion piece in Politico, overtly entering into the political discourse over HFT, is concerning, to say the least. And the Columbia Business School connection can only raise the level of curiosity. Who can forget the image of Professor Jones' colleague Glenn Hubbard squirming and terminating an interview that questioned his earnings from the financial sector in the film Inside Job?
This weekend, I had the opportunity to read the larger article to understand what Professor Jones had accomplished. The Politico piece had touted it as "the first meta-analysis of scholarly work on high-frequency trading." This promised to be an important contribution to knowledge about HFT. A "meta-analysis" is a statistical approach in which the data derived from other studies is combined to yield new conclusions.
Unfortunately, meta-analysis is also subject to very a specific problem that has been raised in the field of medical statistics: "A recent study published in JAMA reviewed 29 meta-analyses from high impact journals and found that conflicts of interests in the studies underlying the meta-analyses were rarely disclosed." The biased researchers cherry pick studies that are simpatico with the goals of their paymasters and are dismissive of the pesky studies that challenge these goals. Could this be the case here?
Reading the professor's work revealed the third hint. It was not a meta-analysis in the sense described above. It was merely a review of studies that are well known to those who have an interest in the literature, similar to the recently published paper I authored. The main difference is that it is far more superficial and does not endeavor to draw new meaning from those studies by integrating actual market realities. As a persuasive device, using the highfalutin term "meta-analysis" did suggest a major statistical research effort in the POLITICO opinion article.
Nonetheless, the article does employ the techniques of cherry picking and dismissiveness. Studies that support HFT are praised and contrary studies are dismissed almost derisively. The article concentrates on conventional concepts of market liquidity and the narrow measure of value, transaction costs. Perhaps the support from Citadel was insufficient to fund deeper analysis.
The HFT business is described as benign and passive, a group of companies seeking to make the market more efficient and taking a fair profit in return. In particular, Jones describes HFT traders as "market makers," traders that quote prices to both buy and sell intending to profit from the spread between the two. The activity involves placing "resting" orders that market participants can rely on to gain comfort that they can obtain execution at a predictable price. To state the obvious, the ability to change tactics at speeds measured in milliseconds is hardly consistent with the reliability of "resting" orders.
Of course, in this business model, the profit potential is subject to geometrically lower profit potential as the margins are squeezed by increasing price efficiency. Nonetheless, the Citadels of the world are engaged in a technology arms race that costs them billions. Perhaps they have lost track of the need to justify these investments with potential profit.
Obviously, this is not true. The benign and passive activity described in the article undoubtedly is a part of HFT. But so are aggressive and disruptive tactics that impair the reliability of markets, costing the economy dearly and transferring value to the users of HFT. This is not the forum for a point-by-point discussion of Professor Jones' article. But it is clear that the superficial and conventional review of the literature cannot begin to explain why the financial sector sees a return on investment to shave fractions of milliseconds off of the speed of information flows. Eric Hunsader, the founder of Nanex, provides the better description of HFT a high speed market date feed service.
In summary, HFT algos reduce the value of resting orders [for example, a market maker's orders] and increase the value of how fast orders can be placed and cancelled. This results in the illusion of liquidity. We can't understand why this is allowed to continue, because at the core, it is pure manipulation.
We can expect the POLITICO article to be trotted around Washington to influence debate. The professor even provides a point-by-point commentary on why specific policies that would injure HFT are bad ideas. Hopefully, the lobbyists' efforts will be viewed with a very large grain of salt.
