Christmas is a time of high stress and, in a lot of people's wallets, low liquidity. Oddly enough, some of the big finance questions of 2014 revolved around high stress and low liquidity. In particular, observers wondered, do high-frequency traders (HFTs) withdraw from equity markets during uncertain and stressful times, causing "phantom liquidity" and putting excess downwards pressure on stock prices? Now, thanks to a paper by Robert Korajczyk of the Kellogg School of Management at Northwestern University and Dermot Murphy of the University of Illinois at Chicago, published by the Investment Industry Regulatory Organization of Canada (IIROC) and using Canadian data, we are a step closer to an answer.
In short, it seems that HFTs do tend to withdraw from the markets at a higher rate than traditional "designated market makers" during times of high stress, which is valuable information for regulators trying to help guide Canada's markets.
I've made my position on HFTs clear in the past but generally I think that they are a good thing – they add liquidity, help with price discovery and generally make markets more efficient. As Professors Korjczyk and Murphy state "the ultimate goal of finance is to facilitate the efficient allocation of capital." In other words, the better markets function, the stronger the real economy will be. I know it sounds like a strong take, but finance is not about your retirement portfolio – even if, broadly, more efficient capital allocations will help grow your portfolio, your interests as an investor come second to market efficiency.
This is not a blanket endorsement of high frequency trading. There are costs. Some of those costs are simple social wastefulness: a mad fast cable from New York City to Chicago may not be the best use of $300-million. Other costs are troublesome for our capital markets. In situations where stocks and markets face downward pressure, there is a fear that HFTs will withdraw quickly, decreasing liquidity and exacerbating declines. Quotes posted by HFTs will disappear when sellers (who are often HFTs themselves) try to sell to HFTs during times of stress, forcing share prices rapidly downwards. Some blame the infamous "Flash Crash" of 2010 or Apple's huge share price drop on Dec. 1 on such behaviour. There are other factors that contribute to such declines as well – stop orders designed to mitigate losses by automatically selling shares at a predefined floor but that can flood the market with supply during a decline.
Similarly, HFTs may post orders that they later withdraw when a large institutional trader attempts to execute on it. While HFTs are merely unthinking algorithms, their programmers have a pretty good reason for withdrawing trades in both situations – machines are dumb and people are smart, and HFT algorithms are wise to avoid big trades which could come from smart, better informed investors. During times of stress, HFTs may be even less willing to enter into large trades. Such decreased liquidity is not necessarily a bad thing – when HFTs notice smart investors trying to sell large blocks of shares, and react by removing liquidity and causing prices to rapidly fall to a lower, but correct price, that's an example of the market being efficient. On the other hand, these events could be seen as negative if they increase volatility.
Contrast this with the story of traditional, designated market makers (who, oddly enough, can also be HFTs), who are bound by Toronto Stock Exchange regulations, among other things, to provide a market for a security within a spread defined by the exchange, maintain activity in the market and fill retail-sized orders when the market has insufficient liquidity. And these two groups are what Profs. Korjczyk and Murphy study.
Unsurprisingly, Profs. Korjczyk and Murphy find that HFTs go from supplying about 25 per cent of liquidity during non-stressful times to around 15 per cent of liquidity during stressful times, while DMMs provide slightly more liquidity during times of stress but still only provide about 2 per cent of the market's liquidity in either period. As expected, HFTs provide less liquidity the most for largest, most aggressive, stressful trades – situations where they are at a particular informational disadvantage.
Profs. Korjczyk and Murphy show that when HFTs are providing liquidity, they lower the cost of a trade – trades executed quickly aren't subject to as much price risk as trades that take longer to trade in a less liquid market. When HFTs withdraw liquidity, trades become more expensive and, therefore, investors face not only the costs of declining prices, but also the costs of decreased liquidity.
This is all a nice Christmas present for regulators. HFTs are here to stay and, in my estimation, help our capital markets. But it's necessary to understand HFT behaviour. The more regulators know about what role HFTs actually play in Canadian markets – and where old fashioned designated market makers still have a job to do, namely in stabilizing stressed markets as HFTs flee – the less likely they are to be swayed by anti-HFT hysteria. Or, for that matter, by opportunistic HFT proponents.
Interestingly, a recent, unpublished paper from three Canadian researchers and an American, using NASDAQ data, finds that HFTs are net-liquidity providers during extreme price jumps and tend to stabilize markets during these situations. While this is somewhat contradictory to the Korjczyk and Murphy study, the different results could be driven by different definitions of stress scenarios – in particular, the unpublished paper may be more likely to catch large stock dumps like stop orders, which contain little information and HFTs would be wise to trade on. Regardless, neither paper suggests that HFTs are a force for catastrophe, even if Professors Korjczyk and Murphy's study suggests that there may be ways for regulators to improve liquidity during times of stress.
So, even if you're facing a personal finance-related liquidity constraint come January, you can take solace that, most likely, our markets won't be – HFTs or not.
Editor's Note: An earlier version of this article misspelled Robert Korajczyk's name.