Stocks quoted in this article:
As we've noted often recently, small-cap volatility has hit record levels versus big-cap volatility this year. "Record" requires a little perspective, as we can only go back to 2007 on Russell 2000 Index (RUT) volatility, but still, it's very noteworthy.
Perhaps we can blame (or thank) high-frequency trading (HFT). This, from Credit Suisse via The Wall Street Journal:
New data out from Credit Suisse points to some of the salutary effects of high-frequency trading. Researchers at the Swiss bank's New York offices posit that HFT activity has helped inoculate large-caps from the effects of macroeconomic market stresses.
Their analysis, in a note released Wednesday, shows price volatility is much more subdued in the stocks of companies with large market capitalizations -- whose deeper liquidity HFT generally prefers -- than it is in their small-cap peers.
Looking at stocks that moved at least 1% within one minute, researchers found spikes in small-caps at moments of instability, like the 2011 U.S. debt downgrade. But in large-caps, there wasn't a blip.
… As Credit Suisse put it: "The numbers have been declining consistently every year post-crisis, as HFT has become a larger part of the market. Not so for smallcaps."
It brings up the eternal question about coincidence vs. causation. Is HFT really causing intraday volatility to compress, or are we just selectively viewing a time period where volatility is declining anyway and adding an arbitrary threshold on top of it?
Credit Suisse uses that 1% within one-minute criteria as a proxy for volatility. And that's fine. But, perhaps, in a generally low-volatility environment, that simply knocks out a host of big caps. Remember, smaller caps are just naturally more volatile than big caps, anyway. Not to mention they, on average, will have much lower absolute-dollar price tags on them. That might matter in a big way in a study like the one mentioned above … it's much more likely for a relatively lightly traded $10 stock to blip a dime in a heartbeat than it is for a heavily traded $100 stock to blip $1. And again, that's especially true in a generally low-volatility backdrop.
What if they reduced the criteria to 0.5%, would that change how their graphs look? I don't know for sure, but I suspect it would.
I don't believe their study itself is flawed in any particular way, I just believe they reach an unprovable conclusion. It needs more corroborating evidence. HFT would reduce volatility if it added liquidity to the marketplace. It certainly adds to volume, but liquidity is not synonymous with volume.
Here's a pretty long read from Nanex highlighting a few milliseconds of trading in Ford Motor Company (NYSE:F). Long story short, it's your classic "Trader tries to lift offer, buy order gets rerouted, offers cancel and/or gets lifted before trader fills order" story that happens over and over again in big-cap names. It's a pretty typical event in a world of algorithms, and it's really tough to make the logical case that this has served to increase liquidity and reduce volatility.
Bottom line is, there's lots of moving parts that go into market volatility. HFT certainly has an impact on volatility, but there's really no single way to isolate that impact short of having a "control" market somewhere that's identical in every other way except for the algos. Big stocks blipping 1% relatively infrequently over a time period when HFT has grown does not in any way prove that HFT keeps stocks from blipping 1%.
Disclaimer: Mr. Warner's opinions expressed above do not necessarily represent the views of Schaeffer's Investment Research.