Over the past two weeks, I've highlighted the impact of High Frequency Trading (HFT) on the stock market, including the 2010 Flash Crash example. Obviously, today HFT makes up a big part of the trading that goes on at Wall Street. In this post, I'd like to consider another ongoing issue involving HFT - what is it's impact on market volatility? Does HFT make the market more stable, less stable, or actually have no effect on volatility?
Last week, I introduced the subject of volatility as the variation of price over time. This is true, both in a specific sense of individual stocks and more generally as you look at the whole market. In a more volatile market, prices fluctuate more frequently and with more magnitude - there's more movement. Lately, especially within the past year, market executives have described the market as more volatile than ever. Their evidence are indices like the S&P volatility index (mentioned last week) that seek to measure the overall market's stability. Can we attribute this increased volatility to HFT?
This is the question that has yet to be answered in a unified way. According to an article by Forbes Magazine, HFT doesn't cause volatility in the market but amplifies it.
"High Frequency Trading feeds on volatility because the computers aim to make a profit on the spread - the difference between the buy price and sell price of the stock."
When this spread widens, the computers then go into overdrive, making more and more trades to generate profits. This, according to Forbes, creates a "snowball" effect, amplifying the market move up or down.
Similarly, another source, the Government Office for Science in the UK, points to self-reinforcing feedback loops that amplify the volatility in the market. Their paper, "The Future of Computer Trading in Financial Markets", identifies three main mechanisms by which may lead to instability in markets: nonlinear sensitivities to change (when a small event can have a big impact on the market), incomplete information (where some firms can have more information than others), and internal "endogenous" risks based on feedback loops within the HFT system. The article names six different types of feedback loops that can contribute to this endogenous risk. While these feedback loops in the HFTs may be well-intentioned, they can amplify internal risks and lead to undesired outcomes when a small change loops back on itself and triggers a bigger change which again loops and so on.
The Government Office for Science paper asserts that there is no direct evidence that HFT has increased stock volatility. Yet, using data from a large sample of firms from 1985 to 2009, Frank Zhang from Yale found that HFT increases stock price volatility. The positive correlation that he found was especially strong for the top 3,000 stocks in the market. However, does correlation prove causation? Not necessarily.
Additionally, we have to remember that there have been volatile events in the market before HFT came along. According to Forbes, the level of volatility in the Great Depression starting in 1929 was only 0.3% lower than that of the Great Recession of 2008-2009.
Perhaps the best explanation we can give is that HFT and volatility are intermingled. This is position of Jonathan Brogaard of Northwestern, who published a paper in September of last year about the subject. Using mostly data from the financial crisis separated into two swathes (HFT activity and non-HFT activity), he ran a statistical method called a Granger Causality Analysis to investigate the relationship between volatility and HFT activity. He found that the relationship goes both ways - volatility can cause HFT activity and HFT activity can cause volatility.
It goes without saying that the relationship between market volatility and HFT is a complex one. With hundreds of thousands of trades per day, all controlled by computer algorithms, it's hard for anyone to find a precise explanation for it. However, we do know that they are related in some form - and that's a start.
Sources:
[1] "Are Computers Bringing Down The Stock Market?", Forbes Magazine, August 15, 2011, http://www.forbes.com/sites/investopedia/2011/08/15/are-computers-bringing-down-the-stock-market/2/
[2] Jonathan Brogaard, "High Frequency Trading and Volatility", Northwestern University Kellogg School of Management, September 8, 2011, http://www.wsuc3m.com/HFT_and_Volatility_Final%20Brogaard.pdf
[3] "The Future of Computer Trading in Financial Markets", UK Government Office for Science, 2011, http://www.bis.gov.uk/assets/foresight/docs/computer-trading/11-1276-the-future-of-computer-trading-in-financial-markets
[4] Frank Zhang, "High-Frequency Trading, Stock Volatility, and Price Discovery", Yale School of Management, December 2010, http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1691679
Most of the reasons that you identify for HFT to increase or cause volatility seem to be things that apply to human stock traders also. For instance, if there is a big spread in the buy and sell price, I would imagine that human stock traders would nonlinearly seek to exploit that.
ReplyDelete