In recent months, a plethora of articles have remarked on elevated levels of market volatility, while simultaneously pointing out that computers generate an ever-growing share of the market’s volume. The implication, or in some cases, outright accusation, is that the latter is the cause of the former. While the recent front-page New York Times article on this topic refrains from outright casuistry, it does follow the typical pattern of using coincidence to suggest causality. Not only is this sub-par journalism, but it is bad statistics — particularly when the numbers themselves can easily be used to draw relevant causal inferences.
The article correctly notes that there has been an increasing incidence, in recent times, of days exhibiting unusually high volatility (measured as days when the close-to-close return, or alternatively, the high-low trading range are large in magnitude). However, in sharp contravention to what is now the conventional wisdom, the very same data used by the article’s authors also reveals that computerized trading (in particular, high-frequency trading), bears absolutely none of the culpability for this.
This is easily seen by comparing the data from 2000-2006 to the data from 2007 onward. The year 2007 is of interest here because it was the year that the modern market structure was adopted, due to a mandate of the Securities and Exchange Commission known as Regulation NMS. In many important respects, the adoption of Reg NMS represents a kind of unofficial birth date of what is now known as high-frequency trading.
Sure enough, a cursory examination of the data reveals that from 2000-2006, the S&P 500 moved an average of 0.83% from one day to the next, whereas the market has been much more volatile since then, to the tune of 1.06% per day. But is computerized trading somehow to blame for this, or is the more obvious explanation — uncertainty brought about the global financial crisis — the correct one? Stated differently, who is to blame for volatility — investors or traders?

