A Market without HFT?

High-frequency trading is, by some accounts, a tax on the financial markets. In equity markets alone, HFTs may gain $5 billion in a typical year at the expense of other market participants, according to estimates from the Bank for International Settlements’ Matteo Aquilina, Chicago Booth’s Eric Budish, and Peter O’Neill at the University of New South Wales. (For more, read “How to calculate how much high-frequency trading costs investors.”) In their study of London Stock Exchange data, they find that about a fifth of trading volume came from HFTs racing to beat others to a trade. The typical race involved a tiny amount of money, but it added up to significant sums.

Doing away with such races would, depending on the measure used, eliminate up to a third of the market’s cost of liquidity, the researchers write. And according to Budish, University of Maryland’s Peter Cramton, and University of Notre Dame’s John Shim (a graduate of Booth’s PhD Program), these races could be prevented by better market design.

Most stock exchanges around the world employ continuous trading, whereby assets trade whenever demand (bids) and supply (offers) match up. Within designated market hours, trades happen somewhat randomly. But Budish, Cramton, and Shim argue that in today’s digital markets, this type of trading favors HFT firms, which invest heavily in technology in order to beat everyone else to a trade.

The researchers instead favor the concept of batch auctions, which they say makes speed irrelevant and levels the playing field. In batch auctions, an exchange collects all orders, then executes them together at a single price where the maximum number of orders can be crossed. Right now, exchanges typically hold auctions only when the market opens and closes, but Budish, Cramton, and Shim point out that doing so is a choice—exchanges could run them at more regular intervals, such as every second, if they wanted to.

Some evidence from Asia and Europe supports the notion that exchange design has a big impact on HFT races. Until 2020, the Taiwan Stock Exchange was the largest exchange in the world that held regular batch auctions throughout the day. After the exchange switched to continuous trading, volume rose, but the increase came from faster traders picking off the prices from slower traders as market conditions or news changed, write University of Adelaide’s Ivan Indriawan, University of the Balearic Islands’ Roberto Pascual, and Wilfrid Laurier University’s Andriy Shkilko. Overall trading costs rose due to the actions of HFTs, according to the study. Moreover, the researchers suggest that the increase in volume is the very reason exchanges tend to support continuous trading, as it means more revenues.

Similarly, Peking University’s Yi-Tsung Lee, Norwegian School of Economics’ Roberto Riccó, and Central University of Finance and Economics’ Kai Wang point out that at the TWSE, “continuous trading increased profits for fast investors and losses for individual investors in mid-cap and small-cap [stocks].”

However, they also find that the switch to continuous trading improved efficiency for those same stocks, and led to significant positive, abnormal returns. They conclude that despite HFT races, continuous trading is a better way to run a market.

Looking at a move in the opposite direction on market design, University of Aberdeen’s Zeyu Zhang and University of Edinburgh’s Gbenga Ibikunle write that regulatory restrictions on dark trading in Europe that went into effect in 2018 led to an increase in frequent batch auctions and are linked to reductions in liquidity and informational efficiency. However, the rise in frequent batch auctions is in turn linked to a fall in adverse selection costs for slower traders—who are susceptible to the speed-based strategies deployed by faster ones.