Back to Portfolio for the Future™

Tick Size and High-Frequency Trading

A lot of hopes have been placed on changes in market tick sizes. In the 1990s there was a big push to reduce the tick sizes of securities, allowing them to get down to one cent or fractions thereof.

In the new millennium came a sense of regret. Observers suspected that the low tick sizes had starved margins, and the low margins had caused a wasting away of some of the valuable infrastructure of the securities world.

The results of experiments with larger minimal tick sizes have been disappointing to some. It is possible that some infrastructure did disappear for the reasons that the advocates of larger ticks posit, but even if so … changing the tick sizes back again doesn’t will that infrastructure back into being.

The O’Hara Thesis

On a related issue some recent papers have discussed the impact that changes in tick sizes may have on the automation of trading and the prevalence of high-velocity algorithms. A study last year by Maureen O’Hara, Gideon Saar, and Zhuo Zhong suggested that larger relative tick size benefits HFT, increasing the “adverse selection coming from increased undercutting of limit orders by informed HFT market makers.”

Notice that the O’Hara thesis is distinct from the older thesis of market structure critics. People like Markus Ferber, a Bavarian politician, have long contended high-frequency trading and low tick sizes are both part of the same process, and that together they weaken the price discovery function of markets, substituting phony liquidity for the real stuff. But the O’Hara thesis is that HFT traders are serving a valuable role, and that lowering of tick sizes is an impediment to their playing that role. Lower tick sizes allow for more intense competition among liquidity providers on prices. But speed is not necessarily a decisive asset in that competition. Higher tick sizes (on this hypothesis) limit the ability to compete on prices, meaning that competition moves to another arena: speed. By definition, the HFT “flash boys” are the ones who win this competition, and they proceed to provide liquidity as the side effect of their victory.  That is the O’Hara thesis.

In the latest paper on the subject, though, two researchers with the Federal Reserve and one graduate student at Columbia University look at tick size changes in the interdealer spot foreign exchange market, with an eye to detecting its consequences for high frequency trading. The study indicates that tick changes do matter to market behaviors, though in a way that differs from the usual hypotheses. Either of them.

A Natural (Double) Experiment from EBS

The authors of “What Makes HFTs Tick?” are Alain Chaboud and Ckara Vega, both of the US Federal Reserve, and Avery Dao, of Columbia University, Graduate School of Arts and Sciences, Department of Economics. They worked with data from EBS, a central limit order book used around the world for currency trading. This was a particularly illuminating source of data on the impact of tick size change, because EBS changed its tick sizes two within a brief compass, creating a natural experiment. In March 2011 it reduced the tick size from a pip to a decimal pip on major currency pairs: such as EUR/USD, USD/JPY, EUR/JPY, and so forth. In September 2012 it reversed that change.

So, as tick size changed, what changed in market behavior? Chaboud et al. addressed this question in connection with the euro-dollar exchange rate. They found what one might call “fact one” and “fact two.” Fact one is that “the changes in tick size in the spot foreign exchange market are not accompanied by substantial changes in market liquidity,” contrary to what one might expect on either the O’Hara model or the Ferber critique. They also found—fact two—that immediately following the decline in tick size in March 2011 the role of the spot market in price discovery dropped relative to the role of the corresponding futures market.

What do these two facts mean? The inference that naturally suggests itself is that the flow of orders coming out of HFT trading is less informative than the long-run equilibrium price after a fall in tick size. What had been a dog, wagging the tail of its derivatives market, becomes the tail—getting wagged.

The subsequent reversion to the larger tick size reversed this effect, restoring informative value to the HFT trades.

This finding suggests further research in other markets and tells us that debates over market structure reforms and their consequences are destined to remain murky.