Complexity sciences are popular these days, especially among the experts in new technologies. The question is, however, are they useful for something? That is, can they predict? Or are they doomed to produce analogies and graphics which are indeed impressive, but which heuristic value is null or almost so?

The search for practical applications: this is the Grail that researchers in the field are actively tracking down. We have seen it in a recent article in which Dirck Helbing attempted to simulate the world.

Another researcher whose name has appeared several times recently is Yaneer Bar-Yam, president of the New England Complex Systems Institute (NECSI). He tries hard to understand how complex systems manifest themselves in the different fields of knowledge: ecology, economy, the same manner he tackles economic crises, we are informed by a recent article in Wired magazine.

Bar-Yam affirms having detected, within the mass of market fluctuation data, a fundamental indicator of crashes. This is a market behavior characterized by a growing tendency to imitate among agents. In general, the selling and the buying equilibrate each other: about 50% of the shares go up when 50% go down. When a crash approaches, the “co-movement” increases. In other words, one of the two movements exceeds significantly the threshold of 50%.

Admitedly, a priori this observation has nothing innovative. It is ages that the economists have noticed that the uncontrolable massive behaviors at the source of a “bubble”, or a “panic” on the contrary, often herald crises. But there are differences between Bar-Yam's work and this common view. First, for this researcher this tendency fits in the long term, and does not merely intervene a few days nor even a few months before the “crash”. Thus, before the crisis of 2008 the “co-movement” kept on regularly increasing for more than four years. This was therefore not a situation where the markets enter a phase of rapid panic due to an instinctive collective reaction. In the illustration below, the top curve shows the development of “co-movements”. The lower the level, the greater the tendency to imitate. We can see that this behavior starts to increase as soon as 2003.

Moreover, the cause of this imitation behavior is not exogenous, i.e. it is not created by external events. On the contrary the phenomenon is endogenous, intrinsic to the movement of markets. For the Bar-Yam team, it is still a “panic” effect, but a “self-induced” panic, which does not directly depend on external events or even on speculative “bubbles”, even though these elements can, of course, contribute to amplification at the start of the crisis. For example, concerning the mini-crash of September 17th 2001 (the first stock exchange day after the World Trade Center attacks) Bar-Yam and his colleagues pointed out in their paper published on arXiv that their research tended to “confirm that this event was not only a reaction to September 11, but that it was largely dependent on the market dynamics”.

This capacity to create novelty without needing to receive information from the external world seems indeed to be a fundamental characteristic of complex systems. Before this work on economy, in 2009 Bar-Yam had published an article about biology which came to rather similar conclusions. According to him, species evolution is susceptible to occuring within a population in a purely intrinsic manner, solely from the play of genetic combinations and random mutations. In other words, there is no need to invoke the necessity for the species to adapt to an environment that would become the mutations' referee, as suggested by the classical theory of natural selection. We can only compare this new view of evolution to that of markets evolving spontaneously toward “catasthophes”, independently from the external world or almost.

Predicting Transformations

The Wired article emphasizes that Bar-Yam's work belongs to a new field called “econophysics”, in other words the application to the social realm of behaviors observed in the material world. In the case of economic crises, we are dealing with a process called “phase transition”, such as in the conversion of water into ice or, on the opposite, into steam. In phase transitions, a slow and almost invisible process suddenly accelerates, resulting in a change of state of the system. We therefore have the impression that the conversion happens “by surprise”. That said, the term econophysics may be new, but the study of these transition phases is a long known characteristic of complex systems. We call this the “self-organized criticality” and this concept has been at the heart of the field for more than two decades.

The complexity theories have largely inspired Nassim Nicholas Taleb, who speaks about their consequences in his famous book Le Cygne Noir [“The Black Swan”] (no relationship with Natalie Portman). Taleb uses the complexity theories to make fun of the economists who assess the risks in “Gaussian” terms, by plotting a “mean” curve where the most extreme events are also the least probable. On the contrary, in a “complex” view of the economy or of the society, “catastrophes”, or “black swans” (which, by the way, can also be positive, as shown by the recent revolutions in the arabic world) have all the chances of happening, and a lot faster than we think.

Bar Yam's theories are therefore not revolutionary, but they certainly represent an additional nail in the coffin of classical economics. This research also has the advantage of relying on a relatively simple calculation (well, simple enough for mathematicians), involving a single parameter. It suggests a way of exploring and anticipating the coming of these “catastrophes”, and could therefore, if confirmed, have practical consequences. It could allow, if not to predict the unpredictible, at least to prepare for it intelligently...

Original article on Le Monde