That is indeed a strange title. But stranger are the ways of stock markets.
While financial analysts can guide you through the daily ups and downs of a stock market, accurate forecast of an imminent crash is still a dream. Just like natural calamities, stock market crashes occur frequently and often have repercussions for the global economy. Experts are now looking at natural disasters for clues to understand economic ones.
An interdisciplinary team of scientists from the Indian Institute of Science (IISc), the National University of Ireland, Galway (NUIG) and ENS Cachan – France, came together to examine if market crashes exhibited the same mathematical early warning signatures as natural calamities. Their investigation reveals interesting answers and suggests improved metrics for forecasting a market crash.
Currently, volatility in stock prices is used as a basic risk indicator. However, the recent financial crises of 2007-2008 that caused global markets to shut down temporarily, reminded experts that this is not enough to prepare for a crash. Are there any other signs that we could watch out for? Whispers of a probable answer came from an unexpected field – ecology.
“There is a lot of interest in exchange of ideas between ecology and economics,” points out Prof. Vishwesha Guttal, a mathematical ecologist at the Centre for Ecological Sciences at IISc, and lead author of the published study. This study sprung forth from his discussions with Dr. Srinivas Raghavendra, an economist at NUI Galway, Ireland, looking at the behaviour of financial markets as complex systems.
Financial markets are suggested to be akin to ecological systems with complex feedback loops and sudden critical transitions, also known as ‘tipping points’. So a stock market crash can be compared to unexpected natural transitions such as the onset of Ice Age, desertification of a fertile area, collapse of local fisheries and so on. In recent years, ecologists have been looking for behavioural clues of complex systems in these natural events.
It turns out that many complex systems in nature exhibit ‘critical slowing down’ behaviour before reaching their tipping point. This means that just before a critical transition, it takes longer for them to recover from small disturbances because their internal stabilizing mechanisms become weak. Hence, the system stays ‘disturbed’ for a longer time than usual. Mathematically, one measures this as an increase in variability and autocorrelation in system parameters.
To test this theory on stock market crashes, Guttal and his team rigorously analysed the daily closing data of three major U.S. (Dow Jones Index (DJI), S&P 500 and NASDAQ) and two European (DAX and FTSE) markets spanning the last century. In all cases, they found that variability did increase prior to every known market crash in history. Which means the financial system does get significantly ‘disturbed’ before a crash. But curiously, there was no increase in the autocorrelation of data.
Autocorrelation indicates how similar the data is across different time samples. This means that, once markets are ‘disturbed’, market recovery happens as usual without a ‘slowing down’. This trend is consistent for all crashes across all markets studied by the team. “Many papers suggest that financial meltdowns are also transitions near tipping points, but here we show that they are not,” settles Guttal.
Then why do markets crash?
“We suggest this is because the system is dominated by high stochasticity,” he explains. Their results indicate that if random disturbances in the market grow stronger with time, they can lead to a financial meltdown even if the market is not close to a tipping point. Variability can therefore be an important statistical indicator in early warning signals (EWS) for market crashes, complementing existing indicators such as volatility.
However, there are two major limitations in predictability of such indicators. One, they don’t tell you when a crash may happen. Two, they only suggest a high probability of a crash. In this detailed study of Dow Jones data, 16 EWS emerged from the variability calculations. Of them, 7 were false alarms. But the good news is that there were no failed alarms – the remaining 9 covered every major crash in American market history.
Mr. Nikunj Goel, an undergraduate physics student who worked with Guttal on this study, has developed a basic web app that provides current trends in markets around the globe, including our very own Bombay Stock Exchange. It also shares analyses on historical meltdowns from their published study. The team hopes to add more features to this website and make it more user-friendly.
Could this study have policy implications? “To build robust policies and corrective measures in the future, we need to understand the origin of randomness that drives market meltdowns. This may arise from complex interactions between financial institutions, market microstructure and individual agent behaviour, all adapting at different time scales.” says the economist co-author, Raghavendra. Deconstructing such mind-bogglingly complex systems is a tall order, but Guttal is positive – “The fact that there has been increasing stochasticity prior to every crash, there is a funny element of determinism in it, right?”
This study was published in the open access journal, PLOS ONE and can be accessed from this link.
About the scientists:
Vishwesha Guttal is an Assistant Professor at the Centre for Ecological Studies at the Indian Institute of Science (IISc), Bangalore, India. Srinivasa Raghavendra is a Lecturer at the J. E. Cairnes School of Business and Economics, National University of Ireland, Galway, Ireland, and co-author of this study. Mr. Nikunj Goel, also a co-author of this study, graduated with BSc (Research), majoring in Physics, from IISc in 2015. He is currently pursuing doctoral studies at Yale University, USA. Mr. Quentin Hoarau, a Masters student from Ecole Normale Supérieure of Cachan, France, also contributed to this study.
This team has built a web app for analyzing current market trends and is now tuning its user interface. A preliminary version of the app can be found here.
This article was developed as a press release for the Science Media Center at Indian Institute of Science, Bangalore, India. The press offices at NUI Galway, Nature India and Deccan Herald went on to cover the story.