Tuesday, April 15, 2014

In a commentary in the Financial Post, Prof. Jeffrey MacIntosh pours cold water on the hype surrounding Michael Lewis' new book Flash Boys, about high-frequency traders ("High frequency talker: Author of ‘Flash Boys’ has succeeded in demonizing an innovation that saves investors $9-billion a year," April 14, 2014).

Read the commentary on the Financial Post website, or below.

Michael Lewis’ book, “Flash Boys” – an ostensible indictment of high frequency traders (HFT) – has transformed Mr. Lewis into the undisputed heavyweight champion of the international talk show circuit. Media personalities around the world, including Canada’s own CBC, have been falling all over themselves to give Mr. Lewis a platform to carve up HFT in public. And, by-and-large, they have been uncritically swallowing his alarmist message that financial markets are “rigged.”

What a shame.

Perhaps the most important datum in this story is that alarmism sells. If Mr. Lewis had written a well-documented, impassioned defence of high frequency trading based on solid empirical evidence, his book would likely be well reviewed, sell a few thousand copies, and then quickly be forgotten. But who needs facts when you can get rich peddling soggy half-truths?

Mr. Lewis portrays high-frequency trading firms as modern-day Pirates of the Caribbean who are sabotaging financial markets and siphoning billions of dollars away from good, honest traders. But consider the following. In 2013, the aggregate profits of all HFT operations in the U.S. were on the order of $1-billion. Consider, though, that in excess of $21-trillion worth of stocks was traded in the same period. That means that HFT profits were about 5/1000 of 1% of the total value of stock traded.

What about HFT revenues? Let’s assume that HFT revenues are an order of magnitude higher than profits (which is well in excess of historical averages). HFT revenues would then constitute about 5/100 of 1% of the total value of stock traded.

Thus, whether we focus on profits or revenues, even if we assume that HFT are the evil predators that Michael Lewis makes them out to be, they are removing far less value from U.S. financial markets than a halfway respectable rounding error. Gee, some rigged market.

By-and-large, however, HFT are not predators. Mr. Lewis has assiduously ignored the impressive body of academic evidence demonstrating that when HFT become active in a given financial market, bid/ask spreads substantially narrow. The bid/ask spread is a primary determinant of trading cost, and the narrowing of spreads has greatly benefitted both retail and institutional investors.

Let’s do a rough (and conservative) estimate of the size of that benefit. There are about 1.75 trillion shares traded each year in the U.S.at an average price of $12 per share, and HFT are involved in about half of those trades. Let’s make the extremely conservative assumption that the presence of HFT reduced the average spread by just $0.01 per share. That would mean that the aggregate benefit to investors from the presence of HFT is on the order of $9-billion per year – nearly an order of magnitude greater than the entire HFT industry profit for 2013.

HFT can offer vastly narrower bid/ask spreads than traditional market makers precisely because they trade so fast. By executing a round trip buy/sell or sell/buy in seconds or even milliseconds, the HFT reduce to a bare minimum the risk that the price of the security will change adversely between the first and second leg of the round trip. The speed of HFT trading thus benefits all traders by enhancing liquidity and lowering trading costs.


As emphasized by Douglas Cumming at York University’s Schulich School on this page recently, Michael Lewis focuses on a single aspect of high frequency trading that seems to have some institutional traders’ knickers in a knot – front running. The assertion is that HFT detects the placing of a block order and trades ahead of it in order to earn a profit on the increase in price caused by the placing of the order.

Let us assume that it is correct to say that this practice increases institutional execution costs. Aside from the fact that any front-running costs are likely to be small compared to the reduced bid/ask spreads resulting from high frequency trading activity, institutional traders are not without various self-help remedies. One common practice is to disguise block trades by breaking them up into smaller bits, which are then sent for execution to different trading venues (usually at different times to further obscure the trading pattern). Another is to execute trades through a “dark pool” where institutional block trades interact anonymously through a computer-run order book that is invisible to outsiders. Interestingly, while dark pools exist in Canada, they have garnered only about 5% of all trading activity. It would appear that most Canadian institutions still prefer to use HFT-friendly public trading markets.

Brad Katsuyama, the Canadian who is the subject of Lewis’ unbridled idolatry, thinks he has another solution – setting up an ATS (alternative trading system) called IEX. Like other ATSs, IEX trades exchange-listed stocks. However, it imposes a speed buffer that is designed to reduce the speed advantage of HFT. It also has dispensed with “maker/taker” pricing – a common feature of other ATSs and something that enhances the incentive of HFT to come out and play.

IEX has been open since October. While it has thus far garnered less than 1/3 of 1% of market volume, only time will tell if it does indeed have an industry-beating formula.

The irony, however, is that whether IEX is successful or not, the fact that it exists makes Lewis’ and Katsuyama’s complaints seem like a tempest in a teapot. Success would demonstrate that there is an easy market fix for institutional traders’ complaints. Failure would suggest that most traders prefer to play in an HFT-friendly sandbox after all. Either way, the sky does not seem destined to fall anytime soon.

And what about the much-vilified co-location? Co-location refers to the practice of HFT renting space for its computers in the same building as a stock trading venue’s trade-generating computer. This reduces communication time to far less than the blink of an eye, giving HFT access to posted bid and ask prices before anyone else.

Or does it? The fact is that you don’t have to be a card-carrying high frequency trader to co-locate. Anyone can do it. So if institutional traders want to neutralize the HFT speed advantage, let them do the same. In fact, some already have.

While much more could be said about the things that Michael Lewis doesn’t say, one in particular is worthy of note. If high frequency traders are such bandits, and are “rigging” the equity markets, then why have so many gone out of business in the past several years? The well-regarded TABB Group estimates that revenues derived from high frequency trading of U.S. equities have declined from approximately $7.2-billion in 2009 to an estimated $1.3-billion in 2014. Many smaller firms have gone bust, or been absorbed by larger firms.

In part, the reduction in HFT revenues derives from lower volatility and trading volumes in financial markets. But revenues have also followed the typical path of a maturing technology industry. At first, when only a few players exploit a given technological innovation, the number of competitors is small, the advantage over traditional incumbents large, and profit margins generous. But as time passes, new entrants come on board, incumbents either upgrade their technology or develop defensive strategies, and profit margins shrink dramatically. This pattern has been at work in the HFT industry, and supra-competitive “rents” have long-since evaporated.

In the end, Michael Lewis’ “Flash Boys” is a triumph of marketing over mastery, a book publishing success and a public service failure. It is a useful reminder to reporters and talk show hosts that before you jump on a bandwagon, you should do a little homework and check up on the soundness of the vehicle. Otherwise, you just might be embarrassed when the wheels fall off.