Real costs of high-frequency trading4 min read . Updated: 31 Aug 2016, 04:32 AM IST
When it comes to speed trading, there is overwhelming evidence that it's been largely beneficial to the market
We all love speed. We want 4G speeds for our phones, faster processors for our computers and BMWs for the road. So, why should it be different when it comes to our trading? Does it matter that some traders can trade a few hundred times before others can even blink? Does it matter that exchanges cater to the needs of these high-speed traders and make a quick buck themselves?
These are some of the questions that the Securities and Exchange Board of India (Sebi) has been pondering over, well, for several years now. And as its recent request to the public (for comments on its proposed solutions) shows, it hasn’t made much headway in answering them. Here is an attempt to set the issues in context before evaluating the specific solutions that Sebi is proposing.
To be fair, Sebi is not alone in this quest. Every other major regulator is grappling with the same issue of how to regulate markets amidst this speed race to the bottom. As Michael Lewis writes in his best-selling book, Flash Boys, traders are willing to spend millions of dollars to get a speed edge of a few milliseconds. Just to put this in perspective, a blink of an eye takes about 300 to 400 milliseconds.
When it comes to speed trading (or high-frequency trading, or simply HFT), there is overwhelming empirical evidence that it has been largely beneficial to the markets. Market quality has improved, not just in one market, but in almost all global markets, including our own, where HFT is prevalent. Prices have become more efficient, costs have gone down and volumes have gone up across all these markets.
Yet, regulators feel the need to rein in HFT because of two main concerns: One has to do with their role during volatile periods, while the other revolves around the unfair advantages that they enjoy. Both these concerns stem from the negative impact that HFT has on other traders, and on the broader market as a whole.
Many studies identify HFT as exacerbating fragile market conditions but not causing them (unless there is manipulation involved). Given this, regulators should address volatile periods through special trade rules rather than tightening HFT regulation in general. Limit up/limit down rules and stock-level circuit breakers lower the possibility of wild price swings without disrupting normal functioning of the market. It is not surprising that such rules remain the only real intervention introduced by the US Securities and Exchange Commission after years of researching HFT.
Yet another concern revolves around HFT’s unfair use of market infrastructure, and its faster access to data. Higher system usage, especially when it generates little value, clearly imposes a disproportionate cost on others. Here too, the need to intervene should be targeted, based on perceived risk of an infrastructure breakdown rather than through bottling HFT altogether. How exchanges develop, maintain and fund their infrastructure should be their concern, provided the regulatory consequences of a breakdown are credible.
Much of the angst against HFT comes from their faster access to data. Co-locating closer and closer to the exchanges’ computers to shave a few milliseconds creates a natural monopoly in market data. Except the closest trader, all others are likely to be trading on data that is stale. It is like Biff Tannen betting on sports events using the Sports Almanac from the future (in the movie Back to the Future).
The desire to get closer stems from the order queuing rule that exchanges follow. Most of them follow price and time priority which guarantees early traders first dip in the order pool if they have the best price. While there is little empirical evidence of the impact of speed race on trader profits, it can induce perverse incentives for exchanges, especially when they are run as for-profit entities. Exchanges can favour one trader over another, and can monetize assets such as data that can be argued to be a public good. Just this reason alone justifies some level of regulatory intervention.
With this in the backdrop, it is easy to evaluate Sebi’s proposed solutions. A few, such as the minimum resting time, providing market summaries (rather than tick-by-tick data) and separate queues for faster and slower traders can be dismissed outright. They cause more harm than good. Forcing orders to stay for a fixed time will be disastrous during volatile times, and is akin to forcing people to stay put when there is fire everywhere. Similarly, providing only part and not full data is like asking people to review a book reading only its abridged version and not the original. Separate queues can distort prices and could cause market breakdowns, especially if slower traders feel that prices have become stale. No serious regulator has considered these options, and Sebi shouldn’t either.
As discussed before, order-to-trade fee is best left for exchanges to imp-ose, while the solution involving frequent batch auctions, though appealing, is more difficult to implement. In fact, the few exchanges with regular batch auctions have moved away to continuous trading, suggesting its lack of appeal with market participants. This leaves us with only randomization proposals—randomizing entry time and introducing random order processing delays are worth considering. Both reduce the importance of time priority, and hence the benefit to beat others on speed alone. This will ensure some parity among traders, reduce perverse incentives for exchanges, and slow down this ‘winner take all’ race to the bottom among HFT firms while preserving most of the benefits of HFT. It is important to test it out with a pilot containing a few stocks before rolling it out to the broader market.
Venkatesh Panchapagesan is adjunct professor of finance & control area at IIM Bangalore.
Comments are welcome at email@example.com