High-Frequency Trading: Tame the Beast, or Will the Beast Devour Itself?
High-frequency trading has been back in the news and amongst the thoughts of policymakers over the past weeks, with the publication of a speech by Bank of England Executive Director Andrew Haldane on the ‘Race to Zero’ and the publication by IOSCO of a consultation paper on the impact of technological changes (such as HFT) on market integrity and efficiency.
Haldane cites a number of developments in the markets in recent years that he believes are at least partly attributable to the growth of HFT. These include a dramatic rise in stock market turnover and corresponding decline in average holding periods; a rise in the number, and fall in the average size, of trades executed; and a sharp rise in both volatility and correlation across stocks. According to Haldane, this volatility/correlation dynamic for U.S. stocks is now at all-time highs. Perhaps more profoundly, stock price ‘abnormality’ — the persistence of fat-tailed returns — is also on the rise.
Haldane proceeds to discuss the systemic risks associated with these developments; namely, the risk of contagion across trading venues and asset classes from price dislocations in a given asset class — propagated by gyrations in the liquidity supplied by HFTs. The ‘flash crash’ is the most obvious such phenomenon.
So what, if anything, should regulators do to address these risks? Haldane suggests three main things. Firstly, electronic market makers should be required to commit to providing liquidity, irrespective of the direction of market movements, in order to lessen the likelihood of liquidity droughts and associated price dislocations. Secondly, a consistent circuit breaker policy should be implemented across trading venues to prevent contagion spreading across exchanges. And thirdly, imposing a minimum resting time for orders, whilst likely to result in wider spreads, would set a speed limit and improve the resiliency of liquidity.
Unintended Consequences
None of these proposals are new. CFA Institute has supported the first two of these proposals in our comment letters and related outreach over the past year. However, the third is more controversial. As I highlighted at the TradeTech conference in London this past April, enforcing a minimum resting time for orders would prohibit investors from rapidly amending orders in response to changing market conditions. Consequently, those orders would be exposed to greater risk of gaming or exploitation by more sophisticated algorithms. Limit orders essentially provide a free trading option to market participants with superior information. A minimum resting time extends that option, thus exposing the submitter of the order to greater risk of adverse market movements. This could discourage investors from submitting displayed limit orders, thereby reducing overall market transparency and liquidity, and increasing the ‘toxicity’ of on-exchange order flow as investors are driven away into the dark markets. This is more likely to harm market integrity and efficiency than to improve it.
Another proposal that has been suggested elsewhere is to place a cap on the ratio of orders submitted to transactions executed. Whilst this may reduce the volume of message traffic (itself a cost to the market ecosystem), a cap may restrict the ability of investors engaging in statistical arbitrage to adjust their quotes in response to changes in the fair value of related securities. This would result in pricing inefficiencies across related securities and financial instruments.
CFA Institute therefore suggests that regulators should focus on risk management — pre-trade risk controls over order flow and some form of harmonised circuit breakers or trading halts across exchanges — rather than trying to intervene in the trading process itself.
The End Game
On a final note, regulators and some investors may take some comfort in the declining fortunes of HFT firms, which have lost a little of their market share and their profitability over the past year. Going forwards, this trend could continue, for two reasons. Firstly, HFT firms risk cannibalising themselves. By crowding out other investors — or specifically less informed investors — they deprive themselves of the type of order flow that they can most profitably trade against. As noted above, the crowding-out effect occurs because investors want to avoid interacting with the type of highly informed or ‘toxic’ order flow represented by HFT. In this respect, the exchanges’ coveting of HFT is perhaps somewhat short-sighted. The more HFT the exchanges attract, the more ‘toxic’ that exchange becomes to other investors, and therefore the less attractive it becomes as a market place. Secondly, the speed of light is the final frontier in the technological arms race. Getting to that point will not only be costly, but moreover, once it is reached, the game (or at least as we know it) is up.
