| Summary: | When examining a sector of the economy, it can be sometimes difficult to identify the relationships between the underlying variables that compose it. Therefore, we developed a causal analysis technique, capable of converting large amounts of data into a directed graph of cause-effect relationships. The main objective of the technique is to locate the attractors associated with the system, that is, the sets of variables toward which the system tends dynamically. This methodology is based not only on General System Theory, but also on the Graph Theory and a discrete version of Chaos Theory. However, when systems have a large number of variables, applying the technique can be a tedious task. We thus implemented the Smarta application, a causal analysis simulator that allows automating this methodology. The software constitutes a reimplementation and continuation of the application already developed by our research group. We conducted a causal analysis of a system extracted from a database of structural statistics of Spanish industrial sector companies between 2008 to 2015 (the data were obtained from Spain's National Institute of Statistics). We focused on the yearly analysis of companies' structural and economic properties, based on 21 proxy variables. Based on the proposed analysis, we attempted to answer the following questions: how were the survey variables causally related? Were there any groups of independent variables within the system? And what trends did the system follow over the 2008-2015 period? The aim was to propose an alternative to classical statistical methods employed until now.
|
|---|
