We are always searching for reasons as to why things happen, or why things are the way they are. Through the understanding of classical mechanics we can predict the direction and the velocity of a billiard ball once hit with a certain force in a given direction. The works of Aristotle, Keppler, Newton and many others have shown that there is an order to the behaviour of objects under forces.
Thus we study the behaviour and qualities of nature and ourselves, in order to obtain insights that allow us to tame environments, harness energy, build civilisations, and form systems such as education, politics or economics. Where we first did not understand or took events as random, we now find patterns and rules that are obeyed time after time, and the understanding has given us a powerful tool – the power of prediction.
This ability to predict has enabled us to build machines, economic and financial models and launch a rocket to the moon. Predicting though, is only as accurate as our understanding of the particular phenomenon under study.
So it is with some consternation that when we look closer, we find that certain things are not predictable. That they exist without order and behave with chaotic inconsistencies. In economics, we discovered the rules of supply and demand and the connection between savings and investments – how orderly were these discoveries! How insightful! However, as time passes, we find with ‘Akerlof’s lemons’ for example, that asymmetric information means that it is not as simple as that. In fact, many other things violate the simple view of the markets, as exhibited by the rise of behavioural economics in the recent times.
The existence of these anomalies in contrast to the straight-forward models complicate our efforts to verify and therefore make valid predictions. The current trend is to divide these studies into orderly and chaotic phenomena, failing to see that there is actually a gradient, a slope between order and chaos, and that the first step to understanding is to identify the gradient and to measure how steep the change is between order and disorder.
Out of the pockets of disorder, certain orderly organisations may arise. Organisations such as the mafia, shadow banking and the black market. Similarly, out of the pockets of order, disorder may also occur with frightening consequence, as Minsky pointed out. And thus we should progress from thinking through the ambitions of dynamic equilibria, and onto viewing everything as ever-evolving systems, open and closed, interlinked or otherwise, with a propensity to change from chaos to order and perhaps back to chaos again, and if an open system, depending on the input into or output from the system.
Until then, the art of predicting relies on good empirical studies. Studies employing methods such as time-series regressions require us to reach certain confidence levels and therefore need as many data points as possible to identify trends and correlations. The problem here is how far back in time should we go for the hypothesis to remain useful? A simple analogy would be akin to measuring the behaviour of a 25 year old by how he behaved when he was 15. That would seem unfair and lead to an incorrect conclusion.
Do we have enough studies to identify the different ranges we should use for various purposes? Would it not be easier if we could divide historical data into developmental stages such as childhood, teenage years, adulthood and so forth? Or into ages such as the stone age and the iron age? Financial markets evolve. Political systems evolve. And certain evolutions bring such a distinct change that we need more studies identifying these points not only to improve results, but also to bring uniformity across studies that build on one another. To reiterate, different ranges might bring about different conclusions.
In addition, borrowing from the principle of Horner’s method, range is a trade-off to accuracy. e.g. I can predict with good accuracy what I’ll do in the next few hours but it will be hard for me to predict what I will be doing exactly one year from now.