A fascinating new study by researchers from the New England Complex Systems Institute (NECSI) proposes a simple explanation for stock market crashes. Even more tantalizingly, it provides a measurable way to predict when and under what conditions they will occur.
Using the components of the Russell 3000 index, they measured the relative number of stocks that move in the same direction. They found that this metric increases to a specific level before a market crash. They call this “mimicry” or co-movement – when a large number of individual issues on the stock market mimic each other and rise and fall together. Not surprisingly, the trend over the past 10 years has been for more and more individual issues to move in the same direction on any given day.
But, among this overarching trend is another. The chart above is showing the variable that measures mimicry (U) in the top panel and the bottom panel shows the annual change of U relative to its standard deviation (from one year earlier averaged over the previous year). When U falls below the second standard deviation, the market experiences a shock. The vertical lines correspond to the 1987 “Black Monday” crash, the 1997 Asian financial crisis, the 2001 (WTC attack on September 11th) and the most recent credit crisis in 2007.
Even when mimicry is important, underlying conditions that imply increased risk can elevate sensitivity and the tendency to mimicry. Underlying conditions in this context may include internal trends such as market bubbles or external factors such as war, or the financial disruptions that preceded the recent market decline. When panic involves collective action, rather than individual response, precursor fluctuations are likely to exist due to a growing sensitivity to real or random disturbances. Our results suggest that self-induced panic is a critical component of both the current fi financial crisis and large single day drops over recent years.The signature we found, the existence of a large probability of co-movement of stocks on any given day, is a measure of systemic risk and vulnerability to self-induced panic.
In other words, they are suggesting that it is the inherent market structure itself and the changes it undergoes that cause a crash to take place, not any external factors or shocks. To any serious student of the market, this is known intuitively.
This idea of “mimicry” or co-movement was pointed out last year by a few research firms like Birinyi Associates: S&P 500 Herding Wreaks Havoc On Breadth Indicators. But the measure used by them in that analysis (correlation) is not exactly the same as that used in this study. Here is an updated chart showing that the correlation has dropped dramatically from last year’s peak:
Source: FT Short View
Below is the full research report titled, “Predicting economic market crises using measures of collective panic” from the England Complex Systems Institute (NECSI):
http://www.scribd.com/doc/48954783/Predicting-Stock-Market-Crises-Using-Measures-of-Collective-Mimicry
tradersnarrative 
Fascinating study. I would assume the computations necessary to arrive at these results are more that I would be able to perform with my limited resources. I would be very interested in monitoring market mimicry over time. Any ideas as to how we might obtain these data in the future?
Richard, you’d be correct in your assumption. Each data point requires hundreds of individual calculations. This is not something that can be calculated on a spreadsheet; you’d have to have a custom built program to get inputs and then do the calculations.