Bad Medicine: regulating high frequency traders won’t address muddling of dark pools
Recently, I read Scott Patterson’s (@pattersonscott) Dark Pools, the somewhat disturbing recent history of high frequency trading, packaged in a superb narrative that is probably the closest to a Bond thriller that the industry will ever get. He eloquently points out the twists and turns that lead to where we are today. Starting with 24 dead white men in tights meeting under a Manhattan-based buttonwood tree in 1792, US stock exchanges are now captive to faceless computer-driven high frequency trading. Supposedly, the mom and pop investor suffers. A number of well-meaning market reform activists echo this opinion. I suspect they are somewhat missing the point, as the anti-competitive behavior of stock market exchanges has a much more significant effect, than any specific tech-driven participant. The problems attributed to high frequency traders are actually just symptoms of exchanges wielding market power to increase their own profits.
- Special data feeds
- Order types
Colocation refers to exchanges’ willingness to sell rack space physically close to its servers, so that HFTs have faster access to the order flow. Rebates refer to special small sweeteners exchanges give to HFTs, in order to make up for assuming their order taker risk. Special data feeds refer to streams of data about current market conditions, more current than they provide anyone else, for an additional fee. Order types refer to special order types, which are ideally suited to use by computers, because of the split microsecond responsiveness required to make use of them. Although each on their own doesn’t amount to that much of a benefit, together they give a computerized wholesale middleman an unfair advantage.
In all of the above cases, the common denominator is not only the high frequency traders, but also the exchanges, like the NYSE. In fact, exchanges are offering unfair advantages to market participants intentionally, for their own material gain. As a result, the exchanges significantly increase their profit margins and earnings. Only a certain type of trader can take advantage of these benefits, yet exchanges are in the unique position of being capable of offering them. It is HFTs who are dependent on the exchanges, and not the other way around.
Back in the “old days”, i.e. the 1990s, there were a number of completely different exchange practices in the US, which the exchanges used as a deterrent for competition, including:
- Trading in fractions, as opposed to decimals
- Allowing human traders to not execute an order if they didn’t want to
- Attempting to force out early computer based traders by forcing the order to be manually entered into their systems
These were all anti-competitive practices meant to limit access to good prices, which strongly affected retail investors-arguably much more so than the so-called additional hidden cost of high frequency trading.
So what has really changed, then?
One set of middlemen for another. We went from human pit traders, who used their sharp elbows to increase spreads as much as they possibly could, to computer-based programs, which attempt to game the technological and regulatory infrastructure of the exchanges. Thus, they pushed most of the human traders out of the market.
Both of these middlemen types are basically two different types of financial wholesalers. They probably have more in common than they’re willing to admit. They’re “in the moving business”. They make more money, if they make more trades, as they try to profit from spreads. It’s no surprise, in that case, that HFTs trade frequently, more frequently than is humanly possible.
High frequency traders are basically arbitrageurs. They buy low, and sell high, albeit in very short time frames. Such activity on it own only makes the market more efficient. It seeks out any source of inefficiency, and successfully removes it.
The environment is where the problem lies. Let’s say you have a leaky pipe under your bathroom sink. If you change the plumbing, the water will flow in a different way. By claiming the problem is a water (trade) flow pattern, when you have a faulty pipe (bad legal and technical infrastructure), you miss the point entirely.
The real problem
The real problem, in my opinion, is the anti-competitive behavior of the exchanges. By anti-competitive behavior, I’m primarily referring to the synthetic barriers to market entry erected by the exchanges, so that their preferred clients profit.
All four of Arnuk’s “four horsemen” are essentially barriers to entry. If high frequency trading really was so profitable and easy, every recent computer science grad would be throwing together algos in his or her parents’ garage. In reality, the “four horsemen” are designed to keep new entrants out, or at least less profitable than the HFTs which the exchanges like.
As a concept, barriers to entry should be familiar to anyone who’s heard of Michael Porter’s 5 forces, a catalog of microeconomic strategies, that businesses can use to increase their profits, typically gaining benefits by reducing competitive pressure.
Over the last 20 years, there have been major changes at exchanges. The most important of these include:
- Merger activity: which lead to increased concentration, i.e. the Herfindal index, e.g. looking at market cap of listed companies
- Going public, i.e.thus increasing pressure on the exchanges to maximize short term profits
- A number of institutional changes, such as Reg MNS
All of these clearly help exchanges to maximize profits according to Porter’s model, yet none of them have to do explicitly with technology. I would argue that these have a much bigger effect on the mom and pop investor than any changes caused by advancements in technology. Exchanges replaced one set of institutional favoritism practices for another, instead of serving as a public utility that they should be.
The side effects we should be feeling
Arguably the externalities (or side effects in non-economist English) brought about by exchanges are significant enough, that they should be considered a public good, just like electricity or water. From a social macro level, exchanges serve to distribute and transfer risk, to help companies to raise capital, and to allocate capital to the most promising resources. Part of this process results in price discovery, where an extremely complex socioeconomic system is collapsed into one number, i.e. the price, so that everyone can make decisions much more effectively.
Allowing exchanges to distort incentives of market participants for their own benefit subverts this whole process. Once there less players compete, margins and fees will increase. These fees and margins will allow exchanges to reap the benefits of their activity, decreasing their positive social effects (positive externalities). These costs may be hidden in a complex technical infrastructure of algorithms, data, and access. I would argue the technology isn’t the most important factor here. Although it’s easy to spin a dystopian vision where the machines are the root cause of all of society’s problems, the problem lies in misaligned incentives.
As Scott Patterson argues eloquently in Dark Pools, Josh Levine the original founder of Island, the first electronic exchange where computer-based high frequency trading arose, championed technology as a solution to the favoritism practiced by exchanges. To be clear, one of the brightest financial technology minds of our generation claimed technology could have solved the problem of special interest groups, but it didn’t, due to non-technical factors. I think he is still correct, assuming that the institutions around trading promoted fairness and transparency.
At the moment, they don’t. As a result, we have massive dark pools. The real problem is not really technology. It’s muddled incentives and conflicts of interest. These conflicts of interest are making our pools dark, not technology.
Arnuk and his compatriots are right, although for the wrong reasons. Drawing attention to the fact that HFTs are the first wave of terminator machines that will eat grandma’s pension for breakfast is not necessarily the only logical conclusion. Your conclusions depend on your assumptions.
The underlying infrastructure (which should function as a utility) is using good old rent-seeking behavior to distort markets. It benefits a different type of middleman than it did 10 years ago, which isn’t really that important. By going after high frequency traders, regulators won’t address the underlying problem, they’ll just look like they’ve been blinded by science.
Related articles and Links
- High-Speed Trading’s Regulatory Roots (cnbc.com)
- Broken Record Alert: High-Frequency Traders Buying Data Ahead of You (blogs.wsj.com)
- Risk and Reward in High Frequency Trading (nakedcapitalism.com)
- A Market Test for High Speed Trading (cnbc.com)
- Dark Pools
[Disclosure: I may get a symbolic referral fee if you buy Dark Pools via my link. It may almost help pay for a mocha, although it might not be enough.]