Economic Complexity Algorithms in Complex Networks: Applications to Economics and Ecology

Economic Complexity (EC) algorithms estimate the fitness and complexity of the nodes of bipartite networks, such as the network of countries and their exported products. In their application to country-product networks, EC algorithms try to shed light on the hidden capabilities of countries. Capabilities represent the intangible assets driving the development and wealth of countries, such as infrastructures and educational systems. We begin by analyzing the linear Economic Complexity Index (ECI) method, for which a country is fit if it exports complex products. Analogously, a product is complex if fit countries export it. We then study the non-linear Economic Fitness Complexity in its non-homogeneous version (NH-EFC). In this case, countries with high fitness export various products, from very simple to very complex. If a non-fit country exports a product, this product has low complexity. The primary outcome of both algorithms is to deliver the list of countries ranked according to their fitness. These algorithms cease to work when the network is not bipartite, even if there is only one “weak” link between two nodes of the same class (e.g., country-country or product-product). Our task has been to generalize them to deal with non-bipartite networks. We show that the NH-EFC is more stable after introducing small non-bipartite perturbations (random uncorrelated noise), i.e., the perturbation leaves the ranking of countries almost unchanged. Eventually, we apply the NH-EFC to study the complexity of the prey-predator ecosystem in Florida Bay and disclose information on the hidden capabilities of the organisms in the system.