Decision Makers usually have an army of consultants that makes a lot of studies about the probable evolution of the economy, however, in the last years, their predictions have not been very precise. In fact, the successes can be explained from the probability of hitting from luck.

It is difficult that someone that gets his incomes from a certain methodology can recognize that his methodology is wrong, but a good consultant must know when it can be applied and when it cannot.

I passed a course on Theory of Systems and Automatic Regulation with honors when I was at the University. This fact let me to understand well the importance and application field of a model to make predictions and to influence the evolution a system following a desired path.

Models are not reality; they are an approximation to it. For instance, the classic theory of systems is based on linear systems, however, most of real systems are not linear. We can use this classic theory in order to control any physical system, if we use techniques of linearization. We usually substitute the real function for the slope (the first derivative). This is a common technique that is totally correct if the displacements near the equilibrium point are infinitesimal. Of course, they usually are not, but in a practical way if the time constant of the system is several orders of magnitude larger than the time constant of the controller, we can accept that kind of approximation.

If we look at the economy models, we can see that there are very far of explaining the current behavior of the economy, because it has changed due to several factors as an extreme increase of complexity and an extreme modification of its “time constant”. With new information technologies, markets today can evolve in very shorter time than a consultant need only in order to get a set of data to analyze. Models can help us to understand some trends, but we should use it extremely carefully if we want to use it in order to make predictions.

Fortunately, there is a way to know if we can make predictions or not. Quantitative complexity measures are related to this factor: The higher the complexity, the lower our capability to make predictions about the evolution of a system.

One of the main functions of decision makers is planning. In a complex world, we must understand that we must superpose a function over this. Planning has no sense if we cannot forecast, then, a new activity has arisen: we need to preserve a healthy value of complexity in order that our planning can provide some value.

This reasoning is valid even if we use non-linear models that we consider more precise, because the behavior of systems can change suddenly. This is common even in physics. All we know that the properties of a rock of frost are different from water, and that the change of phase has an extremely not linear behavior.

I like to use a simple example to explain what can be happen in the current economy that makes current models no sense: Imagine a fluid that is moving following a laminar regime and pass suddenly to a turbulent one. The physical laws that explain its behavior are completely different. If we can take advantage of it to make a plane fly, we need to change the shape of its wings when it overcomes the speed of sound (from cross shape to arrow shape).

Complexity is an intrinsic property of the system and is a totally model-free technique, then it is an analysis tool that can be used in “every phase” of the economy. It can be used to know if we can make predictions, or if we had better to plan actions to improve our resilience.

## Leave a comment

Comments feed for this article